From 58e72f50cc0fd342ffed8be2b21cda7ea2604585 Mon Sep 17 00:00:00 2001 From: Vladimir Mandic Date: Sat, 5 Sep 2026 11:57:53 +0200 Subject: [PATCH] sort reference models Signed-off-by: Vladimir Mandic --- data/reference-base.json | 1543 +++++++++++++++++---------------- data/reference-cloud.json | 18 +- data/reference-community.json | 198 ++--- data/reference-distilled.json | 239 ++--- data/reference-nunchaku.json | 116 +-- data/reference-quantized.json | 202 ++--- 6 files changed, 1165 insertions(+), 1151 deletions(-) diff --git a/data/reference-base.json b/data/reference-base.json index 39f430c3b..34415efe2 100644 --- a/data/reference-base.json +++ b/data/reference-base.json @@ -26,684 +26,21 @@ "desc": "This stable-diffusion-2 model is resumed from stable-diffusion-2-base (512-base-ema.ckpt) and trained for 150k steps using a v-objective on the same dataset. Resumed for another 140k steps on 768x768 images", "extras": "width: 768, height: 768, sampler: DEIS, steps: 20, cfg_scale: 6.0" }, - "StabilityAI StableDiffusion XL": { - "path": "stabilityai/stable-diffusion-xl-base-1.0", - "preview": "stabilityai--stable-diffusion-xl-base-1.0.jpg", - "desc": "Stable Diffusion XL (SDXL) is AI image generation model that is tailored towards more photorealistic outputs with more detailed imagery and composition compared to previous SD models, including SD 2.1. It can make realistic faces and better image composition, all while using shorter and simpler prompts at a greatly increased base resolution of 1024x1024. Just like its predecessors, SDXL has the ability to generate image variations using image-to-image prompting, inpainting (reimagining of the selected parts of an image), and outpainting (creating new parts that lie outside the image borders).", - "variant": "fp16", - "size": 32.0, - "date": "2023 July" - }, - "Microsoft Lens": { - "path": "Jinstudio/Lens", - "preview": "microsoft--Lens.jpg", - "desc": "Microsoft Lens is a text-to-image DiT model using GPT-OSS chat-style prompt encoding and Flux2 VAE decoding.", - "size": 28.43, - "date": "2026 May" - }, - "Boogu Image 0.1 Base": { - "path": "Boogu/Boogu-Image-0.1-Base", - "preview": "Boogu--Boogu-Image-0.1-Base.jpg", - "desc": "Boogu Image 0.1 Base is the core Boogu text-to-image model with Qwen3-VL instruction encoding and a mixed single/double-stream diffusion transformer.", - "size": 35.81, - "date": "2026 June" - }, - "Boogu Image 0.1 Edit": { - "path": "Boogu/Boogu-Image-0.1-Edit", - "preview": "Boogu--Boogu-Image-0.1-Edit.jpg", - "desc": "Boogu Image 0.1 Edit is optimized for image editing with the same instruction-aware Boogu transformer and flow-match scheduler.", - "size": 35.81, - "date": "2026 June" - }, - "Microsoft Lens Base": { - "path": "Jinstudio/Lens-Base", - "preview": "microsoft--Lens-Base.jpg", - "desc": "Microsoft Lens-Base is the base variant of Lens for text-to-image generation with GPT-OSS prompt features.", - "size": 28.43, - "date": "2026 May" - }, - "StabilityAI Stable Cascade": { - "path": "huggingface/stabilityai/stable-cascade", - "variant": "bf16", - "desc": "Stable Cascade is a diffusion model built upon the Würstchen architecture and its main difference to other models like Stable Diffusion is that it is working at a much smaller latent space. Why is this important? The smaller the latent space, the faster you can run inference and the cheaper the training becomes. How small is the latent space? Stable Diffusion uses a compression factor of 8, resulting in a 1024x1024 image being encoded to 128x128. Stable Cascade achieves a compression factor of 42, meaning that it is possible to encode a 1024x1024 image to 24x24, while maintaining crisp reconstructions. The text-conditional model is then trained in the highly compressed latent space. Previous versions of this architecture, achieved a 16x cost reduction over Stable Diffusion 1.5", - "preview": "stabilityai--stable-cascade.jpg", - "extras": "sampler: Default, cfg_scale: 4.0, cfg_image: 1.0", - "size": 2.59, - "date": "2024 February" - }, - "StabilityAI Stable Diffusion 3.0 Medium": { - "path": "stabilityai/stable-diffusion-3-medium-diffusers", - "variant": "fp16", - "desc": "Stable Diffusion 3 Medium is a Multimodal Diffusion Transformer (MMDiT) text-to-image model that features greatly improved performance in image quality, typography, complex prompt understanding, and resource-efficiency", - "preview": "stabilityai--stable-diffusion-3.jpg", - "extras": "sampler: Default, cfg_scale: 7.0", - "size": 28.87, - "date": "2024 June" - }, - "StabilityAI Stable Diffusion 3.5 Medium": { - "path": "stabilityai/stable-diffusion-3.5-medium", - "variant": "fp16", - "desc": "Stable Diffusion 3.5 Medium is a Multimodal Diffusion Transformer with improvements (MMDiT-X) text-to-image model that features improved performance in image quality, typography, complex prompt understanding, and resource-efficiency.", - "preview": "stabilityai--stable-diffusion-3_5-medium.jpg", - "extras": "sampler: Default, cfg_scale: 7.0", - "size": 25.55, - "date": "2024 October" - }, - "StabilityAI Stable Diffusion 3.5 Large": { - "path": "stabilityai/stable-diffusion-3.5-large", - "variant": "fp16", - "desc": "Stable Diffusion 3.5 Large is a Multimodal Diffusion Transformer (MMDiT) text-to-image model that features improved performance in image quality, typography, complex prompt understanding, and resource-efficiency.", - "preview": "stabilityai--stable-diffusion-3_5-large.jpg", - "extras": "sampler: Default, cfg_scale: 7.0", - "size": 36.12, - "date": "2024 October" - }, - "Black Forest Labs FLUX.1 Dev": { - "path": "black-forest-labs/FLUX.1-dev", - "preview": "black-forest-labs--FLUX.1-dev.jpg", - "desc": "FLUX.1 models are based on a hybrid architecture of multimodal and parallel diffusion transformer blocks, scaled to 12B parameters and builing on flow matching", - "extras": "sampler: Default, cfg_scale: 3.5", - "size": 31.42, - "date": "2024 July" - }, - "Black Forest Labs FLUX.1 Schnell": { - "path": "black-forest-labs/FLUX.1-schnell", - "preview": "black-forest-labs--FLUX.1-schnell.jpg", - "desc": "FLUX.1 models are based on a hybrid architecture of multimodal and parallel diffusion transformer blocks, scaled to 12B parameters and builing on flow matching. Trained using latent adversarial diffusion distillation, FLUX.1 [schnell] can generate high-quality images in only 1 to 4 steps", - "extras": "sampler: Default, cfg_scale: 3.5", - "size": 31.41, - "date": "2024 July" - }, - "Black Forest Labs FLUX.1 Kontext Dev": { - "path": "black-forest-labs/FLUX.1-Kontext-dev", - "preview": "black-forest-labs--FLUX.1-Kontext-dev.jpg", - "desc": "FLUX.1 Kontext [dev] is a 12 billion parameter rectified flow transformer capable of editing images based on text instructions.", - "extras": "sampler: Default, cfg_scale: 3.5", - "size": 31.42, - "date": "2025 May" - }, - "Black Forest Labs FLUX.1 Krea Dev": { - "path": "black-forest-labs/FLUX.1-Krea-dev", - "preview": "black-forest-labs--FLUX.1-Krea-dev.jpg", - "desc": "FLUX.1 Krea [dev] is a 12 billion parameter rectified flow transformer capable of generating images from text descriptions.", - "extras": "sampler: Default, cfg_scale: 4.5", - "size": 31.42, - "date": "2025 July" - }, - "Black Forest Labs FLUX.2 Dev": { - "path": "black-forest-labs/FLUX.2-dev", - "preview": "black-forest-labs--FLUX.2-dev.jpg", - "desc": "FLUX.2 generates high-quality images while maintaining character and style consistency across multiple reference images, following structured prompts, reading and writing complex text, adhering to brand guidelines, and reliably handling lighting, layouts, and logos.", - "size": 105.06, - "date": "2025 November" - }, - "Black Forest Labs FLUX.2 Klein Base 4B": { - "path": "black-forest-labs/FLUX.2-klein-base-4B", - "preview": "black-forest-labs--FLUX.2-klein-base-4B.jpg", - "desc": "FLUX.2-klein-base-4B is the undistilled 4 billion parameter base model of FLUX.2-klein. Requires 50 inference steps for full quality but offers flexibility for fine-tuning. Supports text-to-image and multi-reference editing. Apache 2.0 licensed.", - "extras": "sampler: Default, cfg_scale: 4.0, steps: 50", - "size": 14.87, - "date": "2026 January" - }, - "Black Forest Labs FLUX.2 Klein Base 9B": { - "path": "black-forest-labs/FLUX.2-klein-base-9B", - "preview": "black-forest-labs--FLUX.2-klein-base-9B.jpg", - "desc": "FLUX.2-klein-base-9B is the undistilled 9 billion parameter base model of FLUX.2-klein. Requires 50 inference steps for full quality but offers flexibility for fine-tuning. Supports text-to-image and multi-reference editing. Non-commercial license.", - "extras": "sampler: Default, cfg_scale: 4.0, steps: 50", - "size": 32.32, - "date": "2026 January" - }, - "Owen777 UltraFlux-v1": { - "path": "Owen777/UltraFlux-v1", - "preview": "Owen777--UltraFlux-v1.jpg", - "desc": "UltraFlux-v1 is a FLUX.1-dev based text-to-image model optimized for native 4K and multi-aspect-ratio generation with improved composition consistency.", - "extras": "sampler: Default, cfg_scale: 4.0, steps: 50", - "size": 31.58, - "date": "2025 November" - }, - "Z-Image": { - "path": "Tongyi-MAI/Z-Image", - "preview": "Tongyi-MAI--Z-Image.jpg", - "desc": "Z-Image, an efficient image generation foundation model built on a Single-Stream Diffusion Transformer architecture. It preserves the complete training signal with full CFG support, enabling aesthetic versatility from hyper-realistic photography to anime, enhanced output diversity, and robust negative prompting for artifact suppression. Ideal base for LoRA training, ControlNet, and semantic conditioning.", - "extras": "sampler: Default, cfg_scale: 4.0, steps: 50", - "size": 19.11, - "date": "2026 January" - }, - "Z-Image-Turbo": { - "path": "Tongyi-MAI/Z-Image-Turbo", - "preview": "Tongyi-MAI--Z-Image-Turbo.jpg", - "desc": "Z-Image-Turbo, a distilled version of Z-Image that matches or exceeds leading competitors with only 8 NFEs (Number of Function Evaluations). It excels in photorealistic image generation, bilingual text rendering (English & Chinese), and robust instruction adherence.", - "extras": "sampler: Default, cfg_scale: 1.0, steps: 9", - "size": 30.58, - "date": "2025 November" - }, - "Ideogram 4": { - "path": "CalamitousFelicitousness/Ideogram-4-bf16-Diffusers", - "preview": "CalamitousFelicitousness--Ideogram-4-bf16-Diffusers.jpg", - "desc": "Ideogram 4 is Ideogram's first open-weight text-to-image model: a two 9.3B flow-matching DiTs that uses a Qwen3-VL vision-language model as its text encoder, with strong in-image text rendering. Requires structured JSON-caption prompts; prompt-enhance (on by default) rewrites a plain prompt into one.", - "extras": "sampler: Default, cfg_scale: 7.0, steps: 20, width: 1024, height: 1024", - "size": 49.9, - "date": "2026 June" - }, - "Krea 2 Base": { - "path": "CalamitousFelicitousness/Krea-2-Base-Diffusers", - "preview": "CalamitousFelicitousness--Krea-2-Base-Diffusers.jpg", - "desc": "Krea 2 (K2) Base is the undistilled foundation model of the Krea 2 family, trained from scratch by Krea. A 12.9B-parameter single-stream flow-matching DiT that uses a Qwen3-VL-4B vision-language model as its text encoder and the Qwen-Image VAE. The base checkpoint is intended for fine-tuning and LoRA training; LoRAs trained on it apply to Krea 2 Turbo.", - "extras": "sampler: Default, cfg_scale: 4.5, steps: 52", - "size": 33.5, - "date": "2026 June" - }, - "Baidu ERNIE-Image": { - "path": "baidu/ERNIE-Image", - "preview": "baidu--ERNIE-Image.jpg", - "desc": "ERNIE-Image is a text-to-image diffusion transformer model that combines a Mistral3 text encoder with a FlowMatch transformer and Flux2-style VAE for 1024px image generation.", - "extras": "sampler: Default, cfg_scale: 4.0, steps: 50", - "size": 22.29, - "date": "2026 April" - }, - "NucleusAI Nucleus-Image": { - "path": "NucleusAI/Nucleus-Image", - "preview": "NucleusAI--Nucleus-Image.jpg", - "desc": "Nucleus-Image is a sparse mixture-of-experts diffusion transformer with 17B total parameters and about 2B active parameters per forward pass, using a Qwen3-VL text encoder and Qwen-Image VAE for 1024px generation.", - "variant": "bf16", - "extras": "sampler: Default, cfg_scale: 8.0, steps: 50", - "size": 48.09, - "date": "2026 March" - }, - "Qwen-Image": { - "path": "Qwen/Qwen-Image", - "preview": "Qwen--Qwen-Image.jpg", - "desc": "Qwen-Image, an image generation foundation model in the Qwen series that achieves significant advances in complex text rendering and precise image editing.", - "size": 53.74, - "date": "2025 August" - }, - "Qwen-Image-2512": { - "path": "Qwen/Qwen-Image-2512", - "preview": "Qwen--Qwen-Image-2512.jpg", - "desc": "Qwen-Image-2512 is an Qwen Image successor, that significantly reduces the AI-generated look, got finer natural detailils and improved text rendering.", - "size": 53.74, - "date": "2025 December" - }, - "Qwen-Image-Edit": { - "path": "Qwen/Qwen-Image-Edit", - "preview": "Qwen--Qwen-Image-Edit.jpg", - "desc": "Qwen-Image-Edit, the image editing version of Qwen-Image. Built upon our 20B Qwen-Image model, Qwen-Image-Edit successfully extends Qwen-Image’s unique text rendering capabilities to image editing tasks, enabling precise text editing.", - "size": 53.74, - "date": "2025 August" - }, - "Qwen-Image-Edit-2509": { - "path": "Qwen/Qwen-Image-Edit-2509", - "preview": "Qwen--Qwen-Image-Edit-2509.jpg", - "desc": "Qwen-Image-Edit, the image editing version of Qwen-Image. Built upon our 20B Qwen-Image model, Qwen-Image-Edit successfully extends Qwen-Image’s unique text rendering capabilities to image editing tasks, enabling precise text editing.", - "size": 53.74, - "date": "2025 September" - }, - "Qwen-Image-Edit-2511": { - "path": "Qwen/Qwen-Image-Edit-2511", - "preview": "Qwen--Qwen-Image-Edit-2511.jpg", - "desc": "Key enhancements: mitigate image drift, improved character consistency, enhanced industrial design generation, and strengthened geometric reasoning ability.", - "size": 53.74, - "date": "2025 December" - }, - "Qwen-Image-Layered": { - "path": "Qwen/Qwen-Image-Layered", - "preview": "Qwen--Qwen-Image-Layered.jpg", - "desc": "Qwen-Image-Layered, a model capable of decomposing an image into multiple RGBA layers", - "size": 53.74, - "date": "2025 December" - }, - "lodestones Chroma1 HD": { - "path": "lodestones/Chroma1-HD", - "preview": "lodestones--Chroma1-HD.jpg", - "desc": "Chroma is a 8.9B parameter model based on FLUX.1-schnell. It’s fully Apache 2.0 licensed, ensuring that anyone can use, modify, and build on top of it—no corporate gatekeeping. This is the high-res fine-tune of the Chroma1-Base at a 1024x1024 resolution.", - "size": 25.6, - "date": "2025 August" - }, - "lodestones Chroma1 Base": { - "path": "lodestones/Chroma1-Base", - "preview": "lodestones--Chroma1-Base.jpg", - "desc": "Chroma is a 8.9B parameter model based on FLUX.1-schnell. It’s fully Apache 2.0 licensed, ensuring that anyone can use, modify, and build on top of it—no corporate gatekeeping. This is the core 512x512 model. It's a solid, all-around foundation for pretty much any creative project.", - "size": 25.6, - "date": "2025 July" - }, - "lodestones Chroma1 v50 Preview Annealed": { - "path": "vladmandic/chroma-unlocked-v50-annealed", - "preview": "vladmandic--chroma-unlocked-v50-annealed.jpg", - "desc": "Chroma is a 8.9B parameter model based on FLUX.1-schnell. It’s fully Apache 2.0 licensed, ensuring that anyone can use, modify, and build on top of it—no corporate gatekeeping. Re-tweaked variant with extra noise added.", - "size": 25.6, - "date": "2025 August" - }, - "lodestones Zeta-Chroma": { - "path": "lodestones/Zeta-Chroma", - "preview": "lodestones--Zeta-Chroma.jpg", - "desc": "Zeta-Chroma is a pixel-space diffusion transformer image model from lodestones that generates images directly in RGB space using a NextDiT-style architecture.", - "extras": "sampler: Default, cfg_scale: 3.0, steps: 30", - "size": 12.11, - "date": "2025 December" - }, - "Anima 1.0 Base": { - "path": "vladmandic/Anima-1.0-Base", - "preview": "vladmandic--Anima-1.0-Base.jpg", - "desc": "Anima 1.0 Base with extended 1024-resolution training and expanded dataset coverage for less common artists. A 2B parameter anime-focused text-to-image model based on modified Cosmos-Predict-2B with Qwen3-0.6B text encoder, created by CircleStone Labs and Comfy Org.", - "date": "2026 May", - "size": 4.99 - }, - "Anima 1.0 Aesthetic": { - "path": "CalamitousFelicitousness/Anima-1.0-Aesthetic-Diffusers", - "preview": "CalamitousFelicitousness--Anima-1.0-Aesthetic-Diffusers.jpg", - "desc": "Anima 1.0 Aesthetic, fine-tuned for better consistency and a higher quality default art style. A 2B parameter anime-focused text-to-image model based on modified Cosmos-Predict-2B with Qwen3-0.6B text encoder, created by CircleStone Labs and Comfy Org.", - "extras": "sampler: Default, cfg_scale: 4.5, steps: 30", - "date": "2026 July", - "size": 4.99 - }, - "Anima 2.9B Preview v1": { - "path": "CalamitousFelicitousness/Anima-2.9B-Preview-v1-Diffusers", - "preview": "CalamitousFelicitousness--Anima-2.9B-Preview-v1-Diffusers.jpg", - "desc": "Anima 2.9B preview v1 by Gazingstars, a depth-expanded fine-tune of Anima 1.0 Base: the transformer is grown from 28 to 40 layers and trained on 1.7M additional anime and illustration samples with mixed tag and natural-language captions. Same Qwen3-0.6B text encoder, LLM adapter and VAE as Anima 1.0.", - "extras": "sampler: Default, cfg_scale: 4.0, steps: 30", - "date": "2026 August", - "size": 7.32 - }, - "Meituan LongCat Image": { - "path": "meituan-longcat/LongCat-Image", - "preview": "meituan-longcat--LongCat-Image.jpg", - "desc": "Pioneering open-source and bilingual (Chinese-English) foundation model for image generation, designed to address core challenges in multilingual text rendering, photorealism, deployment efficiency, and developer accessibility prevalent in current leading models.", - "size": 27.28, - "date": "2025 December" - }, - "Meituan LongCat Image-Edit": { - "path": "meituan-longcat/LongCat-Image-Edit", - "preview": "meituan-longcat--LongCat-Image-Edit.jpg", - "desc": "Pioneering open-source and bilingual (Chinese-English) foundation model for image generation, designed to address core challenges in multilingual text rendering, photorealism, deployment efficiency, and developer accessibility prevalent in current leading models.", - "size": 27.28, - "date": "2025 December" - }, - "Ostris Flex.2 Preview": { - "path": "ostris/Flex.2-preview", - "preview": "ostris--Flex.2-preview.jpg", - "desc": "Open Source 8B parameter Text to Image Diffusion Model with universal control and inpainting support built in. Early access preview release. The next version of Flex.1-alpha", - "extras": "sampler: Default, cfg_scale: 3.5", - "size": 24.46, - "date": "2025 April" - }, - "Ostris Flex.1 Alpha": { - "path": "ostris/Flex.1-alpha", - "preview": "ostris--Flex.1-alpha.jpg", - "desc": "Flex.1 alpha is a pre-trained base 8 billion parameter rectified flow transformer capable of generating images from text descriptions. It has a similar architecture to FLUX.1-dev, but with fewer double transformer blocks (8 vs 19)", - "extras": "sampler: Default, cfg_scale: 3.5", - "size": 24.46, - "date": "2025 January" - }, - "Wan-AI Wan2.1 1.3B": { - "path": "Wan-AI/Wan2.1-T2V-1.3B-Diffusers", - "preview": "Wan-AI--Wan2.1-T2V-1.3B-Diffusers.jpg", - "desc": "Wan is an advanced and powerful visual generation model developed by Tongyi Lab of Alibaba Group. It can generate videos based on text, images, and other control signals. The Wan2.1 series models are now fully open-source.", + "DeepFloyd IF Medium": { + "path": "DeepFloyd/IF-I-M-v1.0", + "desc": "DeepFloyd-IF is a pixel-based text-to-image triple-cascaded diffusion model, that can generate pictures with new state-of-the-art for photorealism and language understanding. The result is a highly efficient model that outperforms current state-of-the-art models, achieving a zero-shot FID-30K score of 6.66 on the COCO dataset. It is modular and composed of frozen text mode and three pixel cascaded diffusion modules, each designed to generate images of increasing resolution: 64x64, 256x256, and 1024x1024.", + "preview": "DeepFloyd--IF-I-M-v1.0.jpg", "extras": "sampler: Default", - "size": 26.92, - "date": "2025 March" + "size": 75.87, + "date": "2023 March" }, - "Wan-AI Wan2.1 14B": { - "path": "Wan-AI/Wan2.1-T2V-14B-Diffusers", - "preview": "Wan-AI--Wan2.1-T2V-14B-Diffusers.jpg", - "desc": "Wan is an advanced and powerful visual generation model developed by Tongyi Lab of Alibaba Group. It can generate videos based on text, images, and other control signals. The Wan2.1 series models are now fully open-source.", + "DeepFloyd IF Large": { + "path": "DeepFloyd/IF-I-L-v1.0", + "desc": "DeepFloyd-IF is a pixel-based text-to-image triple-cascaded diffusion model, that can generate pictures with new state-of-the-art for photorealism and language understanding. The result is a highly efficient model that outperforms current state-of-the-art models, achieving a zero-shot FID-30K score of 6.66 on the COCO dataset. It is modular and composed of frozen text mode and three pixel cascaded diffusion modules, each designed to generate images of increasing resolution: 64x64, 256x256, and 1024x1024.", + "preview": "DeepFloyd--IF-I-L-v1.0.jpg", "extras": "sampler: Default", - "size": 74.86, - "date": "2025 March" - }, - "Wan-AI Wan2.2 5B": { - "path": "Wan-AI/Wan2.2-TI2V-5B-Diffusers", - "preview": "Wan-AI--Wan2.2-TI2V-5B-Diffusers.jpg", - "desc": "Wan2.2, offering more powerful capabilities, better performance, and superior visual quality. With Wan2.2, we have focused on incorporating the following technical innovations: MoE Architecture, Data Scalling, Cinematic Aesthetics, Efficient High-Definition Hybrid", - "extras": "sampler: Default", - "size": 31.83, - "date": "2025 July" - }, - "Wan-AI Wan2.2 A14B T2I": { - "path": "Wan-AI/Wan2.2-T2V-A14B-Diffusers", - "preview": "Wan-AI--Wan2.2-T2V-A14B-Diffusers.jpg", - "desc": "Wan2.2, offering more powerful capabilities, better performance, and superior visual quality. With Wan2.2, we have focused on incorporating the following technical innovations: MoE Architecture, Data Scalling, Cinematic Aesthetics, Efficient High-Definition Hybrid", - "extras": "sampler: Default", - "size": 117.51, - "date": "2025 July" - }, - "Wan-AI Wan2.2 A14B I2I": { - "path": "Wan-AI/Wan2.2-I2V-A14B-Diffusers", - "preview": "Wan-AI--Wan2.2-T2V-A14B-Diffusers.jpg", - "desc": "Wan2.2, offering more powerful capabilities, better performance, and superior visual quality. With Wan2.2, we have focused on incorporating the following technical innovations: MoE Architecture, Data Scalling, Cinematic Aesthetics, Efficient High-Definition Hybrid", - "extras": "sampler: Default", - "size": 117.52, - "date": "2025 July" - }, - "Wan-AI Wan2.2 14B VACE": { - "path": "linoyts/Wan2.2-VACE-Fun-14B-diffusers", - "preview": "linoyts--Wan2.2-VACE-Fun-14B-diffusers.jpg", - "desc": "Wan2.2, offering more powerful capabilities, better performance, and superior visual quality. With Wan2.2, we have focused on incorporating the following technical innovations: MoE Architecture, Data Scalling, Cinematic Aesthetics, Efficient High-Definition Hybrid", - "extras": "sampler: Default", - "size": 75.64, - "date": "2025 September" - }, - "MiniMaxAI MiniMax-H3": { - "path": "MiniMaxAI/MiniMax-H3", - "preview": "MiniMaxAI--MiniMax-H3.jpg", - "desc": "MiniMax-H3 generates video with synchronized stereo audio in a single denoising pass through a 33B single-stream transformer with a Qwen3-VL conditioner. In image tabs the model runs in experimental still mode, keeping the first frame of a minimal generation.", - "extras": "sampler: Default", - "size": 134.12, - "date": "2026 August" - }, - "MiniMaxAI MiniMax-H3 Ref2VA": { - "path": "MiniMaxAI/MiniMax-H3", - "subfolder": "ref2va", - "preview": "MiniMaxAI--MiniMax-H3.jpg", - "desc": "The omni-reference variant of MiniMax-H3, sharing one repository with the base model as a separate checkpoint partition. Video with synchronized stereo audio is conditioned on reference images for identity and appearance, with reference rows held clean while video rows denoise.", - "extras": "sampler: Default", - "size": 134.12, - "date": "2026 August" - }, - "Freepik F-Lite": { - "path": "Freepik/F-Lite", - "preview": "Freepik--F-Lite.jpg", - "desc": "F Lite is a 10B parameter diffusion model created by Freepik and Fal, trained exclusively on copyright-safe and SFW content. The model was trained on Freepik's internal dataset comprising approximately 80 million copyright-safe images, making it the first publicly available model of this scale trained exclusively on legally compliant and SFW content.", - "extras": "sampler: Default, cfg_scale: 3.5", - "size": 9.03, - "date": "2025 April" - }, - "Freepik F-Lite Texture": { - "path": "Freepik/F-Lite-Texture", - "preview": "Freepik--F-Lite-Texture.jpg", - "desc": "F Lite is a 10B parameter diffusion model created by Freepik and Fal, trained exclusively on copyright-safe and SFW content. The model was trained on Freepik's internal dataset comprising approximately 80 million copyright-safe images, making it the first publicly available model of this scale trained exclusively on legally compliant and SFW content.", - "extras": "sampler: Default, cfg_scale: 3.5", - "size": 9.03, - "date": "2025 April" - }, - "Freepik F-Lite 7B": { - "path": "Freepik/F-Lite-7B", - "preview": "Freepik--F-Lite-7B.jpg", - "desc": "F Lite is a 10B parameter diffusion model created by Freepik and Fal, trained exclusively on copyright-safe and SFW content. The model was trained on Freepik's internal dataset comprising approximately 80 million copyright-safe images, making it the first publicly available model of this scale trained exclusively on legally compliant and SFW content.", - "extras": "sampler: Default, cfg_scale: 3.5", - "size": 9.03, - "date": "2025 May" - }, - "SDXS DreamShaper 512": { - "path": "IDKiro/sdxs-512-dreamshaper", - "preview": "IDKiro--sdxs-512-dreamshaper.jpg", - "desc": "SDXS: Real-Time One-Step Latent Diffusion Models with Image Conditions", - "extras": "width: 512, height: 512, sampler: CMSI, steps: 1, cfg_scale: 0.0", - "size": 1.64, - "date": "2024 April" - }, - "NVLabs Sana 1.5 1.6B 1k": { - "path": "Efficient-Large-Model/SANA1.5_1.6B_1024px_diffusers", - "desc": "Sana is an efficient model with scaling of training-time and inference time techniques. SANA-1.5 delivers: efficient model growth from 1.6B Sana-1.0 model to 4.8B, achieving similar or better performance than training from scratch and saving 60% training cost; efficient model depth pruning, slimming any model size as you want; powerful VLM selection based inference scaling, smaller model+inference scaling > larger model.", - "preview": "Efficient-Large-Model--SANA1.5_1.6B_1024px_diffusers.jpg", - "size": 9.02, - "date": "2025 March" - }, - "NVLabs Sana 1.5 4.8B 1k": { - "path": "Efficient-Large-Model/SANA1.5_4.8B_1024px_diffusers", - "desc": "Sana is an efficient model with scaling of training-time and inference time techniques. SANA-1.5 delivers: efficient model growth from 1.6B Sana-1.0 model to 4.8B, achieving similar or better performance than training from scratch and saving 60% training cost; efficient model depth pruning, slimming any model size as you want; powerful VLM selection based inference scaling, smaller model+inference scaling > larger model.", - "preview": "Efficient-Large-Model--SANA1.5_4.8B_1024px_diffusers.jpg", - "size": 14.83, - "date": "2025 March" - }, - "NVLabs Sana 1.0 1.6B 4k": { - "path": "Efficient-Large-Model/Sana_1600M_4Kpx_BF16_diffusers", - "desc": "Sana is a text-to-image framework that can efficiently generate images up to 4096 × 4096 resolution. Sana can synthesize high-resolution, high-quality images with strong text-image alignment at a remarkably fast speed, deployable on laptop GPU.", - "preview": "Efficient-Large-Model--Sana_1600M_4Kpx_BF16_diffusers.jpg", - "size": 21.03, - "date": "2025 January" - }, - "NVLabs Sana 1.0 1.6B 2k": { - "path": "Efficient-Large-Model/Sana_1600M_2Kpx_BF16_diffusers", - "desc": "Sana is a text-to-image framework that can efficiently generate images up to 4096 × 4096 resolution. Sana can synthesize high-resolution, high-quality images with strong text-image alignment at a remarkably fast speed, deployable on laptop GPU.", - "preview": "Efficient-Large-Model--Sana_1600M_2Kpx_BF16_diffusers.jpg", - "size": 21.03, - "date": "2024 December" - }, - "NVLabs Sana 1.0 1.6B 1k": { - "path": "Efficient-Large-Model/Sana_1600M_1024px_diffusers", - "desc": "Sana is a text-to-image framework that can efficiently generate images up to 4096 × 4096 resolution. Sana can synthesize high-resolution, high-quality images with strong text-image alignment at a remarkably fast speed, deployable on laptop GPU.", - "preview": "Efficient-Large-Model--Sana_1600M_1024px_diffusers.jpg", - "size": 24.02, - "date": "2024 November" - }, - "NVLabs Sana 1.0 0.6B 0.5k": { - "path": "Efficient-Large-Model/Sana_600M_512px_diffusers", - "desc": "Sana is a text-to-image framework that can efficiently generate images up to 4096 × 4096 resolution. Sana can synthesize high-resolution, high-quality images with strong text-image alignment at a remarkably fast speed, deployable on laptop GPU.", - "preview": "Efficient-Large-Model--Sana_600M_512px_diffusers.jpg", - "size": 15.37, - "date": "2024 December" - }, - "nVidia ChronoEdit": { - "path": "nvidia/ChronoEdit-14B-Diffusers", - "preview": "nvidia--ChronoEdit-14B-Diffusers.jpg", - "desc": "ChronoEdit reframes image editing as a video generation task, using input and edited images as start/end frames to leverage pretrained video models with temporal consistency.", - "size": 82.71, - "date": "2025 October" - }, - "nVidia Cosmos-Predict2 T2I 2B": { - "path": "nvidia/Cosmos-Predict2-2B-Text2Image", - "desc": "Cosmos-Predict2: A family of highly performant pre-trained world foundation models purpose-built for generating physics-aware images, videos and world states for physical AI development.", - "preview": "nvidia--Cosmos-Predict2-2B-Text2Image.jpg", - "size": 13.18, - "date": "2025 April" - }, - "nVidia Cosmos-Predict2 T2I 14B": { - "path": "nvidia/Cosmos-Predict2-14B-Text2Image", - "desc": "Cosmos-Predict2: A family of highly performant pre-trained world foundation models purpose-built for generating physics-aware images, videos and world states for physical AI development.", - "preview": "nvidia--Cosmos-Predict2-14B-Text2Image.jpg", - "size": 36.11, - "date": "2025 April" - }, - "X-Omni SFT": { - "path": "X-Omni/X-Omni-SFT", - "desc": "X-Omni: Reinforcement learning makes discrete autoregressive image generative models great again", - "preview": "X-Omni--X-Omni-SFT.jpg", - "size": 23.8, - "date": "2025 August", - "experimental": true - }, - "VectorSpaceLab OmniGen v1": { - "path": "Shitao/OmniGen-v1-diffusers", - "desc": "OmniGen is a unified image generation model that can generate a wide range of images from multi-modal prompts. It is designed to be simple, flexible and easy to use.", - "preview": "Shitao--OmniGen-v1.jpg", - "size": 7.53, - "date": "2024 December" - }, - "VectorSpaceLab OmniGen v2": { - "path": "OmniGen2/OmniGen2", - "desc": "OmniGen2 is a powerful and efficient unified multimodal model. Unlike OmniGen v1, OmniGen2 features two distinct decoding pathways for text and image modalities, utilizing unshared parameters and a decoupled image tokenizer.", - "preview": "OmniGen2--OmniGen2.jpg", - "size": 15.09, - "date": "2025 June" - }, - "AuraFlow 0.3": { - "path": "fal/AuraFlow-v0.3", - "desc": "AuraFlow v0.3 is the fully open-sourced flow-based text-to-image generation model. The model was trained with more compute compared to the previous version, AuraFlow-v0.2. Compared to AuraFlow-v0.2, the model is fine-tuned on more aesthetic datasets and now supports various aspect ratio, (now width and height up to 1536 pixels).", - "preview": "fal--AuraFlow-v0.3.jpg", - "size": 46.1, - "date": "2024 August" - }, - "AuraFlow 0.2": { - "path": "fal/AuraFlow-v0.2", - "desc": "AuraFlow v0.2 is the fully open-sourced largest flow-based text-to-image generation model. The model was trained with more compute compared to the previous version, AuraFlow-v0.1", - "preview": "fal--AuraFlow-v0.2.jpg", - "size": 46.01, - "date": "2024 July" - }, - "Segmind Vega": { - "path": "huggingface/segmind/Segmind-Vega", - "preview": "segmind--Segmind-Vega.jpg", - "desc": "The Segmind-Vega Model is a distilled version of the Stable Diffusion XL (SDXL), offering a remarkable 70% reduction in size and an impressive 100% speedup while retaining high-quality text-to-image generation capabilities. Trained on diverse datasets, including Grit and Midjourney scrape data, it excels at creating a wide range of visual content based on textual prompts. Employing a knowledge distillation strategy, Segmind-Vega leverages the teachings of several expert models, including SDXL, ZavyChromaXL, and JuggernautXL, to combine their strengths and produce compelling visual outputs.", - "variant": "fp16", - "extras": "sampler: Default, cfg_scale: 9.0", - "size": 9.2, - "date": "2023 November" - }, - "Segmind SegMoE SD 4x2": { - "path": "segmind/SegMoE-SD-4x2-v0", - "preview": "segmind--SegMoE-SD-4x2-v0.jpg", - "desc": "SegMoE-SD-4x2-v0 is an untrained Segmind Mixture of Diffusion Experts Model generated using segmoe from 4 Expert SD1.5 models. SegMoE is a powerful framework for dynamically combining Stable Diffusion Models into a Mixture of Experts within minutes without training", - "extras": "width: 512, height: 512, sampler: Default", - "size": 3.04, - "date": "2024 January" - }, - "Segmind SegMoE XL 4x2": { - "path": "segmind/SegMoE-4x2-v0", - "preview": "segmind--SegMoE-4x2-v0.jpg", - "desc": "SegMoE-4x2-v0 is an untrained Segmind Mixture of Diffusion Experts Model generated using segmoe from 4 Expert SDXL models. SegMoE is a powerful framework for dynamically combining Stable Diffusion Models into a Mixture of Experts within minutes without training", - "extras": "sampler: Default", - "size": 15.77, - "date": "2024 January" - }, - "Pixart-α XL 2 Medium": { - "path": "PixArt-alpha/PixArt-XL-2-512x512", - "desc": "PixArt-α is a Transformer-based T2I diffusion model whose image generation quality is competitive with state-of-the-art image generators (e.g., Imagen, SDXL, and even Midjourney), and the training speed markedly surpasses existing large-scale T2I models. Extensive experiments demonstrate that PIXART-α excels in image quality, artistry, and semantic control. It can directly generate 512px images from text prompts within a single sampling process.", - "preview": "PixArt-alpha--PixArt-XL-2-512x512.jpg", - "extras": "width: 512, height: 512, sampler: Default, cfg_scale: 2.0", - "size": 29.36, - "date": "2023 November" - }, - "Pixart-α XL 2 Large": { - "path": "PixArt-alpha/PixArt-XL-2-1024-MS", - "desc": "PixArt-α is a Transformer-based T2I diffusion model whose image generation quality is competitive with state-of-the-art image generators (e.g., Imagen, SDXL, and even Midjourney), and the training speed markedly surpasses existing large-scale T2I models. Extensive experiments demonstrate that PIXART-α excels in image quality, artistry, and semantic control. It can directly generate 1024px images from text prompts within a single sampling process.", - "preview": "PixArt-alpha--PixArt-XL-2-1024-MS.jpg", - "extras": "sampler: Default, cfg_scale: 2.0", - "size": 20.33, - "date": "2023 November" - }, - "Pixart-Σ Small": { - "path": "huggingface/PixArt-alpha/PixArt-Sigma-XL-2-512-MS", - "desc": "PixArt-Σ, a Diffusion Transformer model (DiT) capable of directly generating images at 4K resolution. PixArt-Σ represents a significant advancement over its predecessor, PixArt-α, offering images of markedly higher fidelity and improved alignment with text prompts.", - "preview": "PixArt-alpha--PixArt-Sigma-XL-2-512-MS.jpg", - "extras": "width: 512, height: 512, sampler: Default, cfg_scale: 2.0", - "size": 2.28, - "date": "2024 April" - }, - "Pixart-Σ Medium": { - "path": "huggingface/PixArt-alpha/PixArt-Sigma-XL-2-1024-MS", - "desc": "PixArt-Σ, a Diffusion Transformer model (DiT) capable of directly generating images at 4K resolution. PixArt-Σ represents a significant advancement over its predecessor, PixArt-α, offering images of markedly higher fidelity and improved alignment with text prompts.", - "preview": "PixArt-alpha--PixArt-Sigma-XL-2-1024-MS.jpg", - "extras": "sampler: Default, cfg_scale: 2.0", - "size": 20.33, - "date": "2024 April" - }, - "Pixart-Σ Large": { - "path": "huggingface/PixArt-alpha/PixArt-Sigma-XL-2-2K-MS", - "desc": "PixArt-Σ, a Diffusion Transformer model (DiT) capable of directly generating images at 4K resolution. PixArt-Σ represents a significant advancement over its predecessor, PixArt-α, offering images of markedly higher fidelity and improved alignment with text prompts.", - "preview": "PixArt-alpha--PixArt-Sigma-XL-2-2K-MS.jpg", - "extras": "sampler: Default, cfg_scale: 2.0", - "size": 2.28, - "date": "2024 April" - }, - "Tencent HunyuanImage 2.1": { - "path": "hunyuanvideo-community/HunyuanImage-2.1-Diffusers", - "desc": "HunyuanImage-2.1, a highly efficient text-to-image model that is capable of generating 2K (2048 × 2048) resolution images.", - "preview": "hunyuanvideo-community--HunyuanImage-2.1-Diffusers.jpg", - "size": 49.48, - "date": "2025 September" - }, - "Tencent HunyuanImage 2.1 Refiner": { - "path": "hunyuanvideo-community/HunyuanImage-2.1-Refiner-Diffusers", - "desc": "HunyuanImage-2.1, a highly efficient text-to-image model that is capable of generating 2K (2048 × 2048) resolution images.", - "preview": "hunyuanvideo-community--HunyuanImage-2.1-Diffusers.jpg", - "size": 45.78, - "date": "2025 September" - }, - "Tencent HunyuanDiT 1.2": { - "path": "Tencent-Hunyuan/HunyuanDiT-v1.2-Diffusers", - "desc": "Hunyuan-DiT : A Powerful Multi-Resolution Diffusion Transformer with Fine-Grained Chinese Understanding.", - "preview": "Tencent-Hunyuan--HunyuanDiT-v1.2-Diffusers.jpg", - "extras": "sampler: Default, cfg_scale: 2.0", - "size": 13.43, - "date": "2024 July" - }, - "Tencent HunyuanDiT 1.1": { - "path": "Tencent-Hunyuan/HunyuanDiT-v1.1-Diffusers", - "desc": "Hunyuan-DiT : A Powerful Multi-Resolution Diffusion Transformer with Fine-Grained Chinese Understanding.", - "preview": "Tencent-Hunyuan--HunyuanDiT-v1.1-Diffusers.jpg", - "extras": "sampler: Default, cfg_scale: 2.0", - "size": 13.49, - "date": "2024 June" - }, - "AlphaVLLM Lumina Next SFT": { - "path": "Alpha-VLLM/Lumina-Next-SFT-diffusers", - "desc": "The Lumina-Next-SFT is a Next-DiT model containing 2B parameters and utilizes Gemma-2B as the text encoder, enhanced through high-quality supervised fine-tuning (SFT).", - "preview": "Alpha-VLLM--Lumina-Next-SFT-diffusers.jpg", - "extras": "sampler: Default", - "size": 8.25, - "date": "2024 June" - }, - "AlphaVLLM Lumina 2": { - "path": "Alpha-VLLM/Lumina-Image-2.0", - "desc": "A Unified and Efficient Image Generative Model. Lumina-Image-2.0 is a 2 billion parameter flow-based diffusion transformer capable of generating images from text descriptions.", - "preview": "Alpha-VLLM--Lumina-Image-2.0.jpg", - "extras": "sampler: Default", - "size": 19.77, - "date": "2025 January" - }, - "AlphaVLLM Lumina DiMOO": { - "path": "Alpha-VLLM/Lumina-DiMOO", - "desc": "Lumina-DiMOO is an omni diffusion large language model for multimodal generation and understanding with text-to-image, image editing, and multimodal understanding capabilities.", - "preview": "Alpha-VLLM--Lumina-DiMOO.jpg", - "extras": "sampler: Default", - "size": 0, - "date": "2025 September" - }, - "HiDream-I1 Fast": { - "path": "HiDream-ai/HiDream-I1-Fast", - "desc": "HiDream-I1 is a new open-source image generative foundation model with 17B parameters that achieves state-of-the-art image generation quality within seconds.", - "preview": "HiDream-ai--HiDream-I1-Fast.jpg", - "extras": "sampler: Default", - "size": 43.94, - "date": "2025 April" - }, - "HiDream-I1 Dev": { - "path": "HiDream-ai/HiDream-I1-Dev", - "desc": "HiDream-I1 is a new open-source image generative foundation model with 17B parameters that achieves state-of-the-art image generation quality within seconds.", - "preview": "HiDream-ai--HiDream-I1-Dev.jpg", - "extras": "sampler: Default", - "size": 43.94, - "date": "2025 April" - }, - "HiDream-I1 Full": { - "path": "HiDream-ai/HiDream-I1-Full", - "desc": "HiDream-I1 is a new open-source image generative foundation model with 17B parameters that achieves state-of-the-art image generation quality within seconds.", - "preview": "HiDream-ai--HiDream-I1-Full.jpg", - "extras": "sampler: Default", - "size": 43.94, - "date": "2025 April" - }, - "HiDream-O1 Image": { - "path": "HiDream-ai/HiDream-O1-Image", - "desc": "HiDream-O1-Image is an 8B pixel-level unified transformer model for text-to-image generation, instruction editing, and multi-reference personalization up to 2048x2048.", - "preview": "HiDream-ai--HiDream-O1-Image.jpg", - "extras": "sampler: Default", - "size": 35.2, - "date": "2026 May" - }, - "HiDream-E1 Full": { - "path": "HiDream-ai/HiDream-E1-Full", - "desc": "HiDream-E1 is an image editing model built on HiDream-I1.", - "preview": "HiDream-ai--HiDream-E1-Full.jpg", - "extras": "sampler: Default", - "size": 43.94, - "date": "2025 April" - }, - "HiDream-E1.1": { - "path": "HiDream-ai/HiDream-E1-1", - "desc": "HiDream-E1 is an image editing model built on HiDream-I1.", - "preview": "HiDream-ai--HiDream-E1-1.jpg", - "extras": "sampler: Default", - "size": 43.94, - "date": "2025 July" - }, - "Kwai Kolors": { - "path": "Kwai-Kolors/Kolors-diffusers", - "desc": "Kolors is a large-scale text-to-image generation model based on latent diffusion, developed by the Kuaishou Kolors team. Trained on billions of text-image pairs, Kolors exhibits significant advantages over both open-source and proprietary models in visual quality, complex semantic accuracy, and text rendering for both Chinese and English characters. Furthermore, Kolors supports both Chinese and English inputs", - "preview": "Kwai-Kolors--Kolors-diffusers.jpg", - "extras": "width: 1024, height: 1024", - "size": 16.59, - "date": "2024 July" + "size": 82.18, + "date": "2023 March" }, "Kandinsky 2.1": { "path": "kandinsky-community/kandinsky-2-1", @@ -721,6 +58,53 @@ "size": 9.34, "date": "2023 June" }, + "StabilityAI StableDiffusion XL": { + "path": "stabilityai/stable-diffusion-xl-base-1.0", + "preview": "stabilityai--stable-diffusion-xl-base-1.0.jpg", + "desc": "Stable Diffusion XL (SDXL) is AI image generation model that is tailored towards more photorealistic outputs with more detailed imagery and composition compared to previous SD models, including SD 2.1. It can make realistic faces and better image composition, all while using shorter and simpler prompts at a greatly increased base resolution of 1024x1024. Just like its predecessors, SDXL has the ability to generate image variations using image-to-image prompting, inpainting (reimagining of the selected parts of an image), and outpainting (creating new parts that lie outside the image borders).", + "variant": "fp16", + "size": 32.0, + "date": "2023 July" + }, + "SalesForce BLIP-Diffusion": { + "path": "salesforce/blipdiffusion", + "desc": "BLIP-Diffusion, a new subject-driven image generation model that supports multimodal control which consumes inputs of subject images and text prompts. Unlike other subject-driven generation models, BLIP-Diffusion introduces a new multimodal encoder which is pre-trained to provide subject representation.", + "preview": "salesforce--blipdiffusion.jpg", + "size": 3.97, + "date": "2023 September" + }, + "InstaFlow 0.9B": { + "path": "XCLiu/instaflow_0_9B_from_sd_1_5", + "desc": "InstaFlow is an ultra-fast, one-step image generator that achieves image quality close to Stable Diffusion. This efficiency is made possible through a recent Rectified Flow technique, which trains probability flows with straight trajectories, hence inherently requiring only a single step for fast inference.", + "preview": "XCLiu--instaflow_0_9B_from_sd_1_5.jpg", + "size": 3.97, + "date": "2023 October" + }, + "Segmind Vega": { + "path": "huggingface/segmind/Segmind-Vega", + "preview": "segmind--Segmind-Vega.jpg", + "desc": "The Segmind-Vega Model is a distilled version of the Stable Diffusion XL (SDXL), offering a remarkable 70% reduction in size and an impressive 100% speedup while retaining high-quality text-to-image generation capabilities. Trained on diverse datasets, including Grit and Midjourney scrape data, it excels at creating a wide range of visual content based on textual prompts. Employing a knowledge distillation strategy, Segmind-Vega leverages the teachings of several expert models, including SDXL, ZavyChromaXL, and JuggernautXL, to combine their strengths and produce compelling visual outputs.", + "variant": "fp16", + "extras": "sampler: Default, cfg_scale: 9.0", + "size": 9.2, + "date": "2023 November" + }, + "Pixart-α XL 2 Medium": { + "path": "PixArt-alpha/PixArt-XL-2-512x512", + "desc": "PixArt-α is a Transformer-based T2I diffusion model whose image generation quality is competitive with state-of-the-art image generators (e.g., Imagen, SDXL, and even Midjourney), and the training speed markedly surpasses existing large-scale T2I models. Extensive experiments demonstrate that PIXART-α excels in image quality, artistry, and semantic control. It can directly generate 512px images from text prompts within a single sampling process.", + "preview": "PixArt-alpha--PixArt-XL-2-512x512.jpg", + "extras": "width: 512, height: 512, sampler: Default, cfg_scale: 2.0", + "size": 29.36, + "date": "2023 November" + }, + "Pixart-α XL 2 Large": { + "path": "PixArt-alpha/PixArt-XL-2-1024-MS", + "desc": "PixArt-α is a Transformer-based T2I diffusion model whose image generation quality is competitive with state-of-the-art image generators (e.g., Imagen, SDXL, and even Midjourney), and the training speed markedly surpasses existing large-scale T2I models. Extensive experiments demonstrate that PIXART-α excels in image quality, artistry, and semantic control. It can directly generate 1024px images from text prompts within a single sampling process.", + "preview": "PixArt-alpha--PixArt-XL-2-1024-MS.jpg", + "extras": "sampler: Default, cfg_scale: 2.0", + "size": 20.33, + "date": "2023 November" + }, "Kandinsky 3.0": { "path": "kandinsky-community/kandinsky-3", "desc": "Kandinsky 3.0 is an open-source text-to-image diffusion model built upon the Kandinsky2-x model family. In comparison to its predecessors, Kandinsky 3.0 incorporates more data and specifically related to Russian culture, which allows to generate pictures related to Russin culture. Furthermore, enhancements have been made to the text understanding and visual quality of the model, achieved by increasing the size of the text encoder and Diffusion U-Net models, respectively.", @@ -730,20 +114,6 @@ "size": 25.94, "date": "2023 November" }, - "Kandinsky 5.0 T2I Lite": { - "path": "kandinskylab/Kandinsky-5.0-T2I-Lite-sft-Diffusers", - "desc": "Kandinsky 5.0 Image Lite is a 6B image generation models 1K resulution, high visual quality and strong text-writing", - "preview": "kandinskylab--Kandinsky-5.0-T2I-Lite-sft-Diffusers.jpg", - "size": 30.0, - "date": "2025 November" - }, - "Kandinsky 5.0 I2I Lite": { - "path": "kandinskylab/Kandinsky-5.0-I2I-Lite-sft-Diffusers", - "desc": "Kandinsky 5.0 Image Lite is a 6B image editing models 1K resulution, high visual quality and strong text-writing", - "preview": "kandinskylab--Kandinsky-5.0-T2I-Lite-sft-Diffusers.jpg", - "size": 30.01, - "date": "2025 November" - }, "Playground v1": { "path": "playgroundai/playground-v1", "desc": "Playground v1 is a latent diffusion model that improves the overall HDR quality to get more stunning images.", @@ -776,6 +146,40 @@ "size": 38.77, "date": "2023 December" }, + "Segmind SegMoE SD 4x2": { + "path": "segmind/SegMoE-SD-4x2-v0", + "preview": "segmind--SegMoE-SD-4x2-v0.jpg", + "desc": "SegMoE-SD-4x2-v0 is an untrained Segmind Mixture of Diffusion Experts Model generated using segmoe from 4 Expert SD1.5 models. SegMoE is a powerful framework for dynamically combining Stable Diffusion Models into a Mixture of Experts within minutes without training", + "extras": "width: 512, height: 512, sampler: Default", + "size": 3.04, + "date": "2024 January" + }, + "Segmind SegMoE XL 4x2": { + "path": "segmind/SegMoE-4x2-v0", + "preview": "segmind--SegMoE-4x2-v0.jpg", + "desc": "SegMoE-4x2-v0 is an untrained Segmind Mixture of Diffusion Experts Model generated using segmoe from 4 Expert SDXL models. SegMoE is a powerful framework for dynamically combining Stable Diffusion Models into a Mixture of Experts within minutes without training", + "extras": "sampler: Default", + "size": 15.77, + "date": "2024 January" + }, + "KOALA 700M": { + "path": "huggingface/etri-vilab/koala-700m-llava-cap", + "variant": "fp16", + "desc": "Fast text-to-image model, called KOALA, by compressing SDXL's U-Net and distilling knowledge from SDXL into our model. KOALA-700M can generate a 1024x1024 image in less than 1.5 seconds on an NVIDIA 4090 GPU, which is more than 2x faster than SDXL.", + "preview": "etri-vilab--koala-700m-llava-cap.jpg", + "extras": "sampler: Default", + "size": 12.93, + "date": "2024 January" + }, + "StabilityAI Stable Cascade": { + "path": "huggingface/stabilityai/stable-cascade", + "variant": "bf16", + "desc": "Stable Cascade is a diffusion model built upon the Würstchen architecture and its main difference to other models like Stable Diffusion is that it is working at a much smaller latent space. Why is this important? The smaller the latent space, the faster you can run inference and the cheaper the training becomes. How small is the latent space? Stable Diffusion uses a compression factor of 8, resulting in a 1024x1024 image being encoded to 128x128. Stable Cascade achieves a compression factor of 42, meaning that it is possible to encode a 1024x1024 image to 24x24, while maintaining crisp reconstructions. The text-conditional model is then trained in the highly compressed latent space. Previous versions of this architecture, achieved a 16x cost reduction over Stable Diffusion 1.5", + "preview": "stabilityai--stable-cascade.jpg", + "extras": "sampler: Default, cfg_scale: 4.0, cfg_image: 1.0", + "size": 2.59, + "date": "2024 February" + }, "Playground v2.5": { "path": "playgroundai/playground-v2.5-1024px-aesthetic", "desc": "Playground v2.5 is a diffusion-based text-to-image generative model, and a successor to Playground v2. Playground v2.5 is the state-of-the-art open-source model in aesthetic quality.", @@ -785,12 +189,126 @@ "size": 38.77, "date": "2024 February" }, - "CogView 4": { - "path": "zai-org/CogView4-6B", - "desc": "An innovative cascaded framework that enhances the performance of text-to-image diffusion. CogView is the first model implementing relay diffusion in the realm of text-to-image generation, executing the task by first creating low-resolution images and subsequently applying relay-based super-resolution.", - "preview": "THUDM--CogView4-6B.jpg", - "size": 28.97, - "date": "2025 March" + "SDXS DreamShaper 512": { + "path": "IDKiro/sdxs-512-dreamshaper", + "preview": "IDKiro--sdxs-512-dreamshaper.jpg", + "desc": "SDXS: Real-Time One-Step Latent Diffusion Models with Image Conditions", + "extras": "width: 512, height: 512, sampler: CMSI, steps: 1, cfg_scale: 0.0", + "size": 1.64, + "date": "2024 April" + }, + "Pixart-Σ Small": { + "path": "huggingface/PixArt-alpha/PixArt-Sigma-XL-2-512-MS", + "desc": "PixArt-Σ, a Diffusion Transformer model (DiT) capable of directly generating images at 4K resolution. PixArt-Σ represents a significant advancement over its predecessor, PixArt-α, offering images of markedly higher fidelity and improved alignment with text prompts.", + "preview": "PixArt-alpha--PixArt-Sigma-XL-2-512-MS.jpg", + "extras": "width: 512, height: 512, sampler: Default, cfg_scale: 2.0", + "size": 2.28, + "date": "2024 April" + }, + "Pixart-Σ Medium": { + "path": "huggingface/PixArt-alpha/PixArt-Sigma-XL-2-1024-MS", + "desc": "PixArt-Σ, a Diffusion Transformer model (DiT) capable of directly generating images at 4K resolution. PixArt-Σ represents a significant advancement over its predecessor, PixArt-α, offering images of markedly higher fidelity and improved alignment with text prompts.", + "preview": "PixArt-alpha--PixArt-Sigma-XL-2-1024-MS.jpg", + "extras": "sampler: Default, cfg_scale: 2.0", + "size": 20.33, + "date": "2024 April" + }, + "Pixart-Σ Large": { + "path": "huggingface/PixArt-alpha/PixArt-Sigma-XL-2-2K-MS", + "desc": "PixArt-Σ, a Diffusion Transformer model (DiT) capable of directly generating images at 4K resolution. PixArt-Σ represents a significant advancement over its predecessor, PixArt-α, offering images of markedly higher fidelity and improved alignment with text prompts.", + "preview": "PixArt-alpha--PixArt-Sigma-XL-2-2K-MS.jpg", + "extras": "sampler: Default, cfg_scale: 2.0", + "size": 2.28, + "date": "2024 April" + }, + "StabilityAI Stable Diffusion 3.0 Medium": { + "path": "stabilityai/stable-diffusion-3-medium-diffusers", + "variant": "fp16", + "desc": "Stable Diffusion 3 Medium is a Multimodal Diffusion Transformer (MMDiT) text-to-image model that features greatly improved performance in image quality, typography, complex prompt understanding, and resource-efficiency", + "preview": "stabilityai--stable-diffusion-3.jpg", + "extras": "sampler: Default, cfg_scale: 7.0", + "size": 28.87, + "date": "2024 June" + }, + "Tencent HunyuanDiT 1.1": { + "path": "Tencent-Hunyuan/HunyuanDiT-v1.1-Diffusers", + "desc": "Hunyuan-DiT : A Powerful Multi-Resolution Diffusion Transformer with Fine-Grained Chinese Understanding.", + "preview": "Tencent-Hunyuan--HunyuanDiT-v1.1-Diffusers.jpg", + "extras": "sampler: Default, cfg_scale: 2.0", + "size": 13.49, + "date": "2024 June" + }, + "AlphaVLLM Lumina Next SFT": { + "path": "Alpha-VLLM/Lumina-Next-SFT-diffusers", + "desc": "The Lumina-Next-SFT is a Next-DiT model containing 2B parameters and utilizes Gemma-2B as the text encoder, enhanced through high-quality supervised fine-tuning (SFT).", + "preview": "Alpha-VLLM--Lumina-Next-SFT-diffusers.jpg", + "extras": "sampler: Default", + "size": 8.25, + "date": "2024 June" + }, + "Black Forest Labs FLUX.1 Dev": { + "path": "black-forest-labs/FLUX.1-dev", + "preview": "black-forest-labs--FLUX.1-dev.jpg", + "desc": "FLUX.1 models are based on a hybrid architecture of multimodal and parallel diffusion transformer blocks, scaled to 12B parameters and builing on flow matching", + "extras": "sampler: Default, cfg_scale: 3.5", + "size": 31.42, + "date": "2024 July" + }, + "Black Forest Labs FLUX.1 Schnell": { + "path": "black-forest-labs/FLUX.1-schnell", + "preview": "black-forest-labs--FLUX.1-schnell.jpg", + "desc": "FLUX.1 models are based on a hybrid architecture of multimodal and parallel diffusion transformer blocks, scaled to 12B parameters and builing on flow matching. Trained using latent adversarial diffusion distillation, FLUX.1 [schnell] can generate high-quality images in only 1 to 4 steps", + "extras": "sampler: Default, cfg_scale: 3.5", + "size": 31.41, + "date": "2024 July" + }, + "AuraFlow 0.2": { + "path": "fal/AuraFlow-v0.2", + "desc": "AuraFlow v0.2 is the fully open-sourced largest flow-based text-to-image generation model. The model was trained with more compute compared to the previous version, AuraFlow-v0.1", + "preview": "fal--AuraFlow-v0.2.jpg", + "size": 46.01, + "date": "2024 July" + }, + "Tencent HunyuanDiT 1.2": { + "path": "Tencent-Hunyuan/HunyuanDiT-v1.2-Diffusers", + "desc": "Hunyuan-DiT : A Powerful Multi-Resolution Diffusion Transformer with Fine-Grained Chinese Understanding.", + "preview": "Tencent-Hunyuan--HunyuanDiT-v1.2-Diffusers.jpg", + "extras": "sampler: Default, cfg_scale: 2.0", + "size": 13.43, + "date": "2024 July" + }, + "Kwai Kolors": { + "path": "Kwai-Kolors/Kolors-diffusers", + "desc": "Kolors is a large-scale text-to-image generation model based on latent diffusion, developed by the Kuaishou Kolors team. Trained on billions of text-image pairs, Kolors exhibits significant advantages over both open-source and proprietary models in visual quality, complex semantic accuracy, and text rendering for both Chinese and English characters. Furthermore, Kolors supports both Chinese and English inputs", + "preview": "Kwai-Kolors--Kolors-diffusers.jpg", + "extras": "width: 1024, height: 1024", + "size": 16.59, + "date": "2024 July" + }, + "AuraFlow 0.3": { + "path": "fal/AuraFlow-v0.3", + "desc": "AuraFlow v0.3 is the fully open-sourced flow-based text-to-image generation model. The model was trained with more compute compared to the previous version, AuraFlow-v0.2. Compared to AuraFlow-v0.2, the model is fine-tuned on more aesthetic datasets and now supports various aspect ratio, (now width and height up to 1536 pixels).", + "preview": "fal--AuraFlow-v0.3.jpg", + "size": 46.1, + "date": "2024 August" + }, + "StabilityAI Stable Diffusion 3.5 Medium": { + "path": "stabilityai/stable-diffusion-3.5-medium", + "variant": "fp16", + "desc": "Stable Diffusion 3.5 Medium is a Multimodal Diffusion Transformer with improvements (MMDiT-X) text-to-image model that features improved performance in image quality, typography, complex prompt understanding, and resource-efficiency.", + "preview": "stabilityai--stable-diffusion-3_5-medium.jpg", + "extras": "sampler: Default, cfg_scale: 7.0", + "size": 25.55, + "date": "2024 October" + }, + "StabilityAI Stable Diffusion 3.5 Large": { + "path": "stabilityai/stable-diffusion-3.5-large", + "variant": "fp16", + "desc": "Stable Diffusion 3.5 Large is a Multimodal Diffusion Transformer (MMDiT) text-to-image model that features improved performance in image quality, typography, complex prompt understanding, and resource-efficiency.", + "preview": "stabilityai--stable-diffusion-3_5-large.jpg", + "extras": "sampler: Default, cfg_scale: 7.0", + "size": 36.12, + "date": "2024 October" }, "CogView 3 Plus": { "path": "zai-org/CogView3-Plus-3B", @@ -806,21 +324,276 @@ "size": 7.61, "date": "2024 October" }, - "KOALA 700M": { - "path": "huggingface/etri-vilab/koala-700m-llava-cap", - "variant": "fp16", - "desc": "Fast text-to-image model, called KOALA, by compressing SDXL's U-Net and distilling knowledge from SDXL into our model. KOALA-700M can generate a 1024x1024 image in less than 1.5 seconds on an NVIDIA 4090 GPU, which is more than 2x faster than SDXL.", - "preview": "etri-vilab--koala-700m-llava-cap.jpg", - "extras": "sampler: Default", - "size": 12.93, - "date": "2024 January" + "NVLabs Sana 1.0 1.6B 1k": { + "path": "Efficient-Large-Model/Sana_1600M_1024px_diffusers", + "desc": "Sana is a text-to-image framework that can efficiently generate images up to 4096 × 4096 resolution. Sana can synthesize high-resolution, high-quality images with strong text-image alignment at a remarkably fast speed, deployable on laptop GPU.", + "preview": "Efficient-Large-Model--Sana_1600M_1024px_diffusers.jpg", + "size": 24.02, + "date": "2024 November" }, - "AIDC Ovis-Image 7B": { - "path": "AIDC-AI/Ovis-Image-7B", - "desc": "Built upon Ovis-U1, Ovis-Image is a 7B text-to-image model specifically optimized for high-quality text rendering, designed to operate efficiently under stringent computational constraints.", - "preview": "AIDC-AI--Ovis-Image-7B.jpg", - "size": 20.29, - "date": "2025 November" + "NVLabs Sana 1.0 1.6B 2k": { + "path": "Efficient-Large-Model/Sana_1600M_2Kpx_BF16_diffusers", + "desc": "Sana is a text-to-image framework that can efficiently generate images up to 4096 × 4096 resolution. Sana can synthesize high-resolution, high-quality images with strong text-image alignment at a remarkably fast speed, deployable on laptop GPU.", + "preview": "Efficient-Large-Model--Sana_1600M_2Kpx_BF16_diffusers.jpg", + "size": 21.03, + "date": "2024 December" + }, + "NVLabs Sana 1.0 0.6B 0.5k": { + "path": "Efficient-Large-Model/Sana_600M_512px_diffusers", + "desc": "Sana is a text-to-image framework that can efficiently generate images up to 4096 × 4096 resolution. Sana can synthesize high-resolution, high-quality images with strong text-image alignment at a remarkably fast speed, deployable on laptop GPU.", + "preview": "Efficient-Large-Model--Sana_600M_512px_diffusers.jpg", + "size": 15.37, + "date": "2024 December" + }, + "VectorSpaceLab OmniGen v1": { + "path": "Shitao/OmniGen-v1-diffusers", + "desc": "OmniGen is a unified image generation model that can generate a wide range of images from multi-modal prompts. It is designed to be simple, flexible and easy to use.", + "preview": "Shitao--OmniGen-v1.jpg", + "size": 7.53, + "date": "2024 December" + }, + "Ostris Flex.1 Alpha": { + "path": "ostris/Flex.1-alpha", + "preview": "ostris--Flex.1-alpha.jpg", + "desc": "Flex.1 alpha is a pre-trained base 8 billion parameter rectified flow transformer capable of generating images from text descriptions. It has a similar architecture to FLUX.1-dev, but with fewer double transformer blocks (8 vs 19)", + "extras": "sampler: Default, cfg_scale: 3.5", + "size": 24.46, + "date": "2025 January" + }, + "NVLabs Sana 1.0 1.6B 4k": { + "path": "Efficient-Large-Model/Sana_1600M_4Kpx_BF16_diffusers", + "desc": "Sana is a text-to-image framework that can efficiently generate images up to 4096 × 4096 resolution. Sana can synthesize high-resolution, high-quality images with strong text-image alignment at a remarkably fast speed, deployable on laptop GPU.", + "preview": "Efficient-Large-Model--Sana_1600M_4Kpx_BF16_diffusers.jpg", + "size": 21.03, + "date": "2025 January" + }, + "AlphaVLLM Lumina 2": { + "path": "Alpha-VLLM/Lumina-Image-2.0", + "desc": "A Unified and Efficient Image Generative Model. Lumina-Image-2.0 is a 2 billion parameter flow-based diffusion transformer capable of generating images from text descriptions.", + "preview": "Alpha-VLLM--Lumina-Image-2.0.jpg", + "extras": "sampler: Default", + "size": 19.77, + "date": "2025 January" + }, + "Wan-AI Wan2.1 1.3B": { + "path": "Wan-AI/Wan2.1-T2V-1.3B-Diffusers", + "preview": "Wan-AI--Wan2.1-T2V-1.3B-Diffusers.jpg", + "desc": "Wan is an advanced and powerful visual generation model developed by Tongyi Lab of Alibaba Group. It can generate videos based on text, images, and other control signals. The Wan2.1 series models are now fully open-source.", + "extras": "sampler: Default", + "size": 26.92, + "date": "2025 March" + }, + "Wan-AI Wan2.1 14B": { + "path": "Wan-AI/Wan2.1-T2V-14B-Diffusers", + "preview": "Wan-AI--Wan2.1-T2V-14B-Diffusers.jpg", + "desc": "Wan is an advanced and powerful visual generation model developed by Tongyi Lab of Alibaba Group. It can generate videos based on text, images, and other control signals. The Wan2.1 series models are now fully open-source.", + "extras": "sampler: Default", + "size": 74.86, + "date": "2025 March" + }, + "NVLabs Sana 1.5 1.6B 1k": { + "path": "Efficient-Large-Model/SANA1.5_1.6B_1024px_diffusers", + "desc": "Sana is an efficient model with scaling of training-time and inference time techniques. SANA-1.5 delivers: efficient model growth from 1.6B Sana-1.0 model to 4.8B, achieving similar or better performance than training from scratch and saving 60% training cost; efficient model depth pruning, slimming any model size as you want; powerful VLM selection based inference scaling, smaller model+inference scaling > larger model.", + "preview": "Efficient-Large-Model--SANA1.5_1.6B_1024px_diffusers.jpg", + "size": 9.02, + "date": "2025 March" + }, + "NVLabs Sana 1.5 4.8B 1k": { + "path": "Efficient-Large-Model/SANA1.5_4.8B_1024px_diffusers", + "desc": "Sana is an efficient model with scaling of training-time and inference time techniques. SANA-1.5 delivers: efficient model growth from 1.6B Sana-1.0 model to 4.8B, achieving similar or better performance than training from scratch and saving 60% training cost; efficient model depth pruning, slimming any model size as you want; powerful VLM selection based inference scaling, smaller model+inference scaling > larger model.", + "preview": "Efficient-Large-Model--SANA1.5_4.8B_1024px_diffusers.jpg", + "size": 14.83, + "date": "2025 March" + }, + "CogView 4": { + "path": "zai-org/CogView4-6B", + "desc": "An innovative cascaded framework that enhances the performance of text-to-image diffusion. CogView is the first model implementing relay diffusion in the realm of text-to-image generation, executing the task by first creating low-resolution images and subsequently applying relay-based super-resolution.", + "preview": "THUDM--CogView4-6B.jpg", + "size": 28.97, + "date": "2025 March" + }, + "Ostris Flex.2 Preview": { + "path": "ostris/Flex.2-preview", + "preview": "ostris--Flex.2-preview.jpg", + "desc": "Open Source 8B parameter Text to Image Diffusion Model with universal control and inpainting support built in. Early access preview release. The next version of Flex.1-alpha", + "extras": "sampler: Default, cfg_scale: 3.5", + "size": 24.46, + "date": "2025 April" + }, + "Freepik F-Lite": { + "path": "Freepik/F-Lite", + "preview": "Freepik--F-Lite.jpg", + "desc": "F Lite is a 10B parameter diffusion model created by Freepik and Fal, trained exclusively on copyright-safe and SFW content. The model was trained on Freepik's internal dataset comprising approximately 80 million copyright-safe images, making it the first publicly available model of this scale trained exclusively on legally compliant and SFW content.", + "extras": "sampler: Default, cfg_scale: 3.5", + "size": 9.03, + "date": "2025 April" + }, + "Freepik F-Lite Texture": { + "path": "Freepik/F-Lite-Texture", + "preview": "Freepik--F-Lite-Texture.jpg", + "desc": "F Lite is a 10B parameter diffusion model created by Freepik and Fal, trained exclusively on copyright-safe and SFW content. The model was trained on Freepik's internal dataset comprising approximately 80 million copyright-safe images, making it the first publicly available model of this scale trained exclusively on legally compliant and SFW content.", + "extras": "sampler: Default, cfg_scale: 3.5", + "size": 9.03, + "date": "2025 April" + }, + "nVidia Cosmos-Predict2 T2I 2B": { + "path": "nvidia/Cosmos-Predict2-2B-Text2Image", + "desc": "Cosmos-Predict2: A family of highly performant pre-trained world foundation models purpose-built for generating physics-aware images, videos and world states for physical AI development.", + "preview": "nvidia--Cosmos-Predict2-2B-Text2Image.jpg", + "size": 13.18, + "date": "2025 April" + }, + "nVidia Cosmos-Predict2 T2I 14B": { + "path": "nvidia/Cosmos-Predict2-14B-Text2Image", + "desc": "Cosmos-Predict2: A family of highly performant pre-trained world foundation models purpose-built for generating physics-aware images, videos and world states for physical AI development.", + "preview": "nvidia--Cosmos-Predict2-14B-Text2Image.jpg", + "size": 36.11, + "date": "2025 April" + }, + "HiDream-I1 Fast": { + "path": "HiDream-ai/HiDream-I1-Fast", + "desc": "HiDream-I1 is a new open-source image generative foundation model with 17B parameters that achieves state-of-the-art image generation quality within seconds.", + "preview": "HiDream-ai--HiDream-I1-Fast.jpg", + "extras": "sampler: Default", + "size": 43.94, + "date": "2025 April" + }, + "HiDream-I1 Dev": { + "path": "HiDream-ai/HiDream-I1-Dev", + "desc": "HiDream-I1 is a new open-source image generative foundation model with 17B parameters that achieves state-of-the-art image generation quality within seconds.", + "preview": "HiDream-ai--HiDream-I1-Dev.jpg", + "extras": "sampler: Default", + "size": 43.94, + "date": "2025 April" + }, + "HiDream-I1 Full": { + "path": "HiDream-ai/HiDream-I1-Full", + "desc": "HiDream-I1 is a new open-source image generative foundation model with 17B parameters that achieves state-of-the-art image generation quality within seconds.", + "preview": "HiDream-ai--HiDream-I1-Full.jpg", + "extras": "sampler: Default", + "size": 43.94, + "date": "2025 April" + }, + "HiDream-E1 Full": { + "path": "HiDream-ai/HiDream-E1-Full", + "desc": "HiDream-E1 is an image editing model built on HiDream-I1.", + "preview": "HiDream-ai--HiDream-E1-Full.jpg", + "extras": "sampler: Default", + "size": 43.94, + "date": "2025 April" + }, + "Black Forest Labs FLUX.1 Kontext Dev": { + "path": "black-forest-labs/FLUX.1-Kontext-dev", + "preview": "black-forest-labs--FLUX.1-Kontext-dev.jpg", + "desc": "FLUX.1 Kontext [dev] is a 12 billion parameter rectified flow transformer capable of editing images based on text instructions.", + "extras": "sampler: Default, cfg_scale: 3.5", + "size": 31.42, + "date": "2025 May" + }, + "Freepik F-Lite 7B": { + "path": "Freepik/F-Lite-7B", + "preview": "Freepik--F-Lite-7B.jpg", + "desc": "F Lite is a 10B parameter diffusion model created by Freepik and Fal, trained exclusively on copyright-safe and SFW content. The model was trained on Freepik's internal dataset comprising approximately 80 million copyright-safe images, making it the first publicly available model of this scale trained exclusively on legally compliant and SFW content.", + "extras": "sampler: Default, cfg_scale: 3.5", + "size": 9.03, + "date": "2025 May" + }, + "VectorSpaceLab OmniGen v2": { + "path": "OmniGen2/OmniGen2", + "desc": "OmniGen2 is a powerful and efficient unified multimodal model. Unlike OmniGen v1, OmniGen2 features two distinct decoding pathways for text and image modalities, utilizing unshared parameters and a decoupled image tokenizer.", + "preview": "OmniGen2--OmniGen2.jpg", + "size": 15.09, + "date": "2025 June" + }, + "Black Forest Labs FLUX.1 Krea Dev": { + "path": "black-forest-labs/FLUX.1-Krea-dev", + "preview": "black-forest-labs--FLUX.1-Krea-dev.jpg", + "desc": "FLUX.1 Krea [dev] is a 12 billion parameter rectified flow transformer capable of generating images from text descriptions.", + "extras": "sampler: Default, cfg_scale: 4.5", + "size": 31.42, + "date": "2025 July" + }, + "lodestones Chroma1 Base": { + "path": "lodestones/Chroma1-Base", + "preview": "lodestones--Chroma1-Base.jpg", + "desc": "Chroma is a 8.9B parameter model based on FLUX.1-schnell. It’s fully Apache 2.0 licensed, ensuring that anyone can use, modify, and build on top of it—no corporate gatekeeping. This is the core 512x512 model. It's a solid, all-around foundation for pretty much any creative project.", + "size": 25.6, + "date": "2025 July" + }, + "Wan-AI Wan2.2 5B": { + "path": "Wan-AI/Wan2.2-TI2V-5B-Diffusers", + "preview": "Wan-AI--Wan2.2-TI2V-5B-Diffusers.jpg", + "desc": "Wan2.2, offering more powerful capabilities, better performance, and superior visual quality. With Wan2.2, we have focused on incorporating the following technical innovations: MoE Architecture, Data Scalling, Cinematic Aesthetics, Efficient High-Definition Hybrid", + "extras": "sampler: Default", + "size": 31.83, + "date": "2025 July" + }, + "Wan-AI Wan2.2 A14B T2I": { + "path": "Wan-AI/Wan2.2-T2V-A14B-Diffusers", + "preview": "Wan-AI--Wan2.2-T2V-A14B-Diffusers.jpg", + "desc": "Wan2.2, offering more powerful capabilities, better performance, and superior visual quality. With Wan2.2, we have focused on incorporating the following technical innovations: MoE Architecture, Data Scalling, Cinematic Aesthetics, Efficient High-Definition Hybrid", + "extras": "sampler: Default", + "size": 117.51, + "date": "2025 July" + }, + "Wan-AI Wan2.2 A14B I2I": { + "path": "Wan-AI/Wan2.2-I2V-A14B-Diffusers", + "preview": "Wan-AI--Wan2.2-T2V-A14B-Diffusers.jpg", + "desc": "Wan2.2, offering more powerful capabilities, better performance, and superior visual quality. With Wan2.2, we have focused on incorporating the following technical innovations: MoE Architecture, Data Scalling, Cinematic Aesthetics, Efficient High-Definition Hybrid", + "extras": "sampler: Default", + "size": 117.52, + "date": "2025 July" + }, + "HiDream-E1.1": { + "path": "HiDream-ai/HiDream-E1-1", + "desc": "HiDream-E1 is an image editing model built on HiDream-I1.", + "preview": "HiDream-ai--HiDream-E1-1.jpg", + "extras": "sampler: Default", + "size": 43.94, + "date": "2025 July" + }, + "Bria 3.2": { + "path": "SahilCarterr/BRIA-3.2", + "desc": "Bria 3.2 is the next-generation commercial-ready text-to-image model. With just 4 billion parameters, it provides exceptional aesthetics and text rendering, evaluated to provide on par results to leading open-source models, and outperforming other licensed models.", + "preview": "briaai--BRIA-3.2.jpg", + "size": 18.11, + "date": "2025 July" + }, + "Qwen-Image": { + "path": "Qwen/Qwen-Image", + "preview": "Qwen--Qwen-Image.jpg", + "desc": "Qwen-Image, an image generation foundation model in the Qwen series that achieves significant advances in complex text rendering and precise image editing.", + "size": 53.74, + "date": "2025 August" + }, + "Qwen-Image-Edit": { + "path": "Qwen/Qwen-Image-Edit", + "preview": "Qwen--Qwen-Image-Edit.jpg", + "desc": "Qwen-Image-Edit, the image editing version of Qwen-Image. Built upon our 20B Qwen-Image model, Qwen-Image-Edit successfully extends Qwen-Image’s unique text rendering capabilities to image editing tasks, enabling precise text editing.", + "size": 53.74, + "date": "2025 August" + }, + "lodestones Chroma1 HD": { + "path": "lodestones/Chroma1-HD", + "preview": "lodestones--Chroma1-HD.jpg", + "desc": "Chroma is a 8.9B parameter model based on FLUX.1-schnell. It’s fully Apache 2.0 licensed, ensuring that anyone can use, modify, and build on top of it—no corporate gatekeeping. This is the high-res fine-tune of the Chroma1-Base at a 1024x1024 resolution.", + "size": 25.6, + "date": "2025 August" + }, + "lodestones Chroma1 v50 Preview Annealed": { + "path": "vladmandic/chroma-unlocked-v50-annealed", + "preview": "vladmandic--chroma-unlocked-v50-annealed.jpg", + "desc": "Chroma is a 8.9B parameter model based on FLUX.1-schnell. It’s fully Apache 2.0 licensed, ensuring that anyone can use, modify, and build on top of it—no corporate gatekeeping. Re-tweaked variant with extra noise added.", + "size": 25.6, + "date": "2025 August" + }, + "X-Omni SFT": { + "path": "X-Omni/X-Omni-SFT", + "desc": "X-Omni: Reinforcement learning makes discrete autoregressive image generative models great again", + "preview": "X-Omni--X-Omni-SFT.jpg", + "size": 23.8, + "date": "2025 August", + "experimental": true }, "HDM-XUT 340M Anime": { "path": "KBlueLeaf/HDM-xut-340M-anime", @@ -829,35 +602,109 @@ "size": 2.2, "date": "2025 August" }, - "SalesForce BLIP-Diffusion": { - "path": "salesforce/blipdiffusion", - "desc": "BLIP-Diffusion, a new subject-driven image generation model that supports multimodal control which consumes inputs of subject images and text prompts. Unlike other subject-driven generation models, BLIP-Diffusion introduces a new multimodal encoder which is pre-trained to provide subject representation.", - "preview": "salesforce--blipdiffusion.jpg", - "size": 3.97, - "date": "2023 September" - }, - "InstaFlow 0.9B": { - "path": "XCLiu/instaflow_0_9B_from_sd_1_5", - "desc": "InstaFlow is an ultra-fast, one-step image generator that achieves image quality close to Stable Diffusion. This efficiency is made possible through a recent Rectified Flow technique, which trains probability flows with straight trajectories, hence inherently requiring only a single step for fast inference.", - "preview": "XCLiu--instaflow_0_9B_from_sd_1_5.jpg", - "size": 3.97, - "date": "2023 October" - }, - "DeepFloyd IF Medium": { - "path": "DeepFloyd/IF-I-M-v1.0", - "desc": "DeepFloyd-IF is a pixel-based text-to-image triple-cascaded diffusion model, that can generate pictures with new state-of-the-art for photorealism and language understanding. The result is a highly efficient model that outperforms current state-of-the-art models, achieving a zero-shot FID-30K score of 6.66 on the COCO dataset. It is modular and composed of frozen text mode and three pixel cascaded diffusion modules, each designed to generate images of increasing resolution: 64x64, 256x256, and 1024x1024.", - "preview": "DeepFloyd--IF-I-M-v1.0.jpg", + "StepFun Step1X-Edit v1.1": { + "path": "stepfun-ai/Step1X-Edit-v1p1-diffusers", + "preview": "stepfun-ai--Step1X-Edit-v1p1-diffusers.jpg", + "desc": "Multimodal image editing model using Step1X transformer architecture with Qwen2.5-VL text encoding, trained with Flow Matching scheduler for high-quality in-context image edits and refinements.", "extras": "sampler: Default", - "size": 75.87, - "date": "2023 March" + "size": 38.91, + "date": "2025 August" }, - "DeepFloyd IF Large": { - "path": "DeepFloyd/IF-I-L-v1.0", - "desc": "DeepFloyd-IF is a pixel-based text-to-image triple-cascaded diffusion model, that can generate pictures with new state-of-the-art for photorealism and language understanding. The result is a highly efficient model that outperforms current state-of-the-art models, achieving a zero-shot FID-30K score of 6.66 on the COCO dataset. It is modular and composed of frozen text mode and three pixel cascaded diffusion modules, each designed to generate images of increasing resolution: 64x64, 256x256, and 1024x1024.", - "preview": "DeepFloyd--IF-I-L-v1.0.jpg", + "Qwen-Image-Edit-2509": { + "path": "Qwen/Qwen-Image-Edit-2509", + "preview": "Qwen--Qwen-Image-Edit-2509.jpg", + "desc": "Qwen-Image-Edit, the image editing version of Qwen-Image. Built upon our 20B Qwen-Image model, Qwen-Image-Edit successfully extends Qwen-Image’s unique text rendering capabilities to image editing tasks, enabling precise text editing.", + "size": 53.74, + "date": "2025 September" + }, + "Wan-AI Wan2.2 14B VACE": { + "path": "linoyts/Wan2.2-VACE-Fun-14B-diffusers", + "preview": "linoyts--Wan2.2-VACE-Fun-14B-diffusers.jpg", + "desc": "Wan2.2, offering more powerful capabilities, better performance, and superior visual quality. With Wan2.2, we have focused on incorporating the following technical innovations: MoE Architecture, Data Scalling, Cinematic Aesthetics, Efficient High-Definition Hybrid", "extras": "sampler: Default", - "size": 82.18, - "date": "2023 March" + "size": 75.64, + "date": "2025 September" + }, + "Tencent HunyuanImage 2.1": { + "path": "hunyuanvideo-community/HunyuanImage-2.1-Diffusers", + "desc": "HunyuanImage-2.1, a highly efficient text-to-image model that is capable of generating 2K (2048 × 2048) resolution images.", + "preview": "hunyuanvideo-community--HunyuanImage-2.1-Diffusers.jpg", + "size": 49.48, + "date": "2025 September" + }, + "Tencent HunyuanImage 2.1 Refiner": { + "path": "hunyuanvideo-community/HunyuanImage-2.1-Refiner-Diffusers", + "desc": "HunyuanImage-2.1, a highly efficient text-to-image model that is capable of generating 2K (2048 × 2048) resolution images.", + "preview": "hunyuanvideo-community--HunyuanImage-2.1-Diffusers.jpg", + "size": 45.78, + "date": "2025 September" + }, + "AlphaVLLM Lumina DiMOO": { + "path": "Alpha-VLLM/Lumina-DiMOO", + "desc": "Lumina-DiMOO is an omni diffusion large language model for multimodal generation and understanding with text-to-image, image editing, and multimodal understanding capabilities.", + "preview": "Alpha-VLLM--Lumina-DiMOO.jpg", + "extras": "sampler: Default", + "size": 0, + "date": "2025 September" + }, + "nVidia ChronoEdit": { + "path": "nvidia/ChronoEdit-14B-Diffusers", + "preview": "nvidia--ChronoEdit-14B-Diffusers.jpg", + "desc": "ChronoEdit reframes image editing as a video generation task, using input and edited images as start/end frames to leverage pretrained video models with temporal consistency.", + "size": 82.71, + "date": "2025 October" + }, + "Bria FIBO": { + "path": "briaai/FIBO", + "preview": "briaai--FIBO.jpg", + "desc": "BRIA FIBO is an 8-billion parameter text-to-image diffusion model using Flow Matching and featuring a lightweight SmolLM3-3B text encoder. Delivers high-quality, detailed image generation with efficient inference.", + "extras": "sampler: Default, cfg_scale: 3.5", + "size": 23.79, + "date": "2025 October" + }, + "Black Forest Labs FLUX.2 Dev": { + "path": "black-forest-labs/FLUX.2-dev", + "preview": "black-forest-labs--FLUX.2-dev.jpg", + "desc": "FLUX.2 generates high-quality images while maintaining character and style consistency across multiple reference images, following structured prompts, reading and writing complex text, adhering to brand guidelines, and reliably handling lighting, layouts, and logos.", + "size": 105.06, + "date": "2025 November" + }, + "Owen777 UltraFlux-v1": { + "path": "Owen777/UltraFlux-v1", + "preview": "Owen777--UltraFlux-v1.jpg", + "desc": "UltraFlux-v1 is a FLUX.1-dev based text-to-image model optimized for native 4K and multi-aspect-ratio generation with improved composition consistency.", + "extras": "sampler: Default, cfg_scale: 4.0, steps: 50", + "size": 31.58, + "date": "2025 November" + }, + "Z-Image-Turbo": { + "path": "Tongyi-MAI/Z-Image-Turbo", + "preview": "Tongyi-MAI--Z-Image-Turbo.jpg", + "desc": "Z-Image-Turbo, a distilled version of Z-Image that matches or exceeds leading competitors with only 8 NFEs (Number of Function Evaluations). It excels in photorealistic image generation, bilingual text rendering (English & Chinese), and robust instruction adherence.", + "extras": "sampler: Default, cfg_scale: 1.0, steps: 9", + "size": 30.58, + "date": "2025 November" + }, + "Kandinsky 5.0 T2I Lite": { + "path": "kandinskylab/Kandinsky-5.0-T2I-Lite-sft-Diffusers", + "desc": "Kandinsky 5.0 Image Lite is a 6B image generation models 1K resulution, high visual quality and strong text-writing", + "preview": "kandinskylab--Kandinsky-5.0-T2I-Lite-sft-Diffusers.jpg", + "size": 30.0, + "date": "2025 November" + }, + "Kandinsky 5.0 I2I Lite": { + "path": "kandinskylab/Kandinsky-5.0-I2I-Lite-sft-Diffusers", + "desc": "Kandinsky 5.0 Image Lite is a 6B image editing models 1K resulution, high visual quality and strong text-writing", + "preview": "kandinskylab--Kandinsky-5.0-T2I-Lite-sft-Diffusers.jpg", + "size": 30.01, + "date": "2025 November" + }, + "AIDC Ovis-Image 7B": { + "path": "AIDC-AI/Ovis-Image-7B", + "desc": "Built upon Ovis-U1, Ovis-Image is a 7B text-to-image model specifically optimized for high-quality text rendering, designed to operate efficiently under stringent computational constraints.", + "preview": "AIDC-AI--Ovis-Image-7B.jpg", + "size": 20.29, + "date": "2025 November" }, "Photoroom PRX 1024": { "path": "Photoroom/prx-1024-t2i-beta", @@ -866,12 +713,72 @@ "size": 19.28, "date": "2025 November" }, - "Photoroom PRXPixel": { - "path": "Photoroom/prxpixel-t2i", - "desc": "PRXPixel is a 7-billion-parameter pixel-space text-to-image model that denoises RGB images directly using a Qwen3-VL text encoder and flow-matching scheduler.", - "preview": "Photoroom--prxpixel-t2i.jpg", - "size": 16.25, - "date": "2026 June" + "Qwen-Image-2512": { + "path": "Qwen/Qwen-Image-2512", + "preview": "Qwen--Qwen-Image-2512.jpg", + "desc": "Qwen-Image-2512 is an Qwen Image successor, that significantly reduces the AI-generated look, got finer natural detailils and improved text rendering.", + "size": 53.74, + "date": "2025 December" + }, + "Qwen-Image-Edit-2511": { + "path": "Qwen/Qwen-Image-Edit-2511", + "preview": "Qwen--Qwen-Image-Edit-2511.jpg", + "desc": "Key enhancements: mitigate image drift, improved character consistency, enhanced industrial design generation, and strengthened geometric reasoning ability.", + "size": 53.74, + "date": "2025 December" + }, + "Qwen-Image-Layered": { + "path": "Qwen/Qwen-Image-Layered", + "preview": "Qwen--Qwen-Image-Layered.jpg", + "desc": "Qwen-Image-Layered, a model capable of decomposing an image into multiple RGBA layers", + "size": 53.74, + "date": "2025 December" + }, + "lodestones Zeta-Chroma": { + "path": "lodestones/Zeta-Chroma", + "preview": "lodestones--Zeta-Chroma.jpg", + "desc": "Zeta-Chroma is a pixel-space diffusion transformer image model from lodestones that generates images directly in RGB space using a NextDiT-style architecture.", + "extras": "sampler: Default, cfg_scale: 3.0, steps: 30", + "size": 12.11, + "date": "2025 December" + }, + "Meituan LongCat Image": { + "path": "meituan-longcat/LongCat-Image", + "preview": "meituan-longcat--LongCat-Image.jpg", + "desc": "Pioneering open-source and bilingual (Chinese-English) foundation model for image generation, designed to address core challenges in multilingual text rendering, photorealism, deployment efficiency, and developer accessibility prevalent in current leading models.", + "size": 27.28, + "date": "2025 December" + }, + "Meituan LongCat Image-Edit": { + "path": "meituan-longcat/LongCat-Image-Edit", + "preview": "meituan-longcat--LongCat-Image-Edit.jpg", + "desc": "Pioneering open-source and bilingual (Chinese-English) foundation model for image generation, designed to address core challenges in multilingual text rendering, photorealism, deployment efficiency, and developer accessibility prevalent in current leading models.", + "size": 27.28, + "date": "2025 December" + }, + "Black Forest Labs FLUX.2 Klein Base 4B": { + "path": "black-forest-labs/FLUX.2-klein-base-4B", + "preview": "black-forest-labs--FLUX.2-klein-base-4B.jpg", + "desc": "FLUX.2-klein-base-4B is the undistilled 4 billion parameter base model of FLUX.2-klein. Requires 50 inference steps for full quality but offers flexibility for fine-tuning. Supports text-to-image and multi-reference editing. Apache 2.0 licensed.", + "extras": "sampler: Default, cfg_scale: 4.0, steps: 50", + "size": 14.87, + "date": "2026 January" + }, + "Black Forest Labs FLUX.2 Klein Base 9B": { + "path": "black-forest-labs/FLUX.2-klein-base-9B", + "preview": "black-forest-labs--FLUX.2-klein-base-9B.jpg", + "desc": "FLUX.2-klein-base-9B is the undistilled 9 billion parameter base model of FLUX.2-klein. Requires 50 inference steps for full quality but offers flexibility for fine-tuning. Supports text-to-image and multi-reference editing. Non-commercial license.", + "extras": "sampler: Default, cfg_scale: 4.0, steps: 50", + "size": 32.32, + "date": "2026 January" + }, + "Z-Image": { + "path": "Tongyi-MAI/Z-Image", + "preview": "Tongyi-MAI--Z-Image.jpg", + "desc": "Z-Image, an efficient image generation foundation model built on a Single-Stream Diffusion Transformer architecture. It preserves the complete training signal with full CFG support, enabling aesthetic versatility from hyper-realistic photography to anime, enhanced output diversity, and robust negative prompting for artifact suppression. Ideal base for LoRA training, ControlNet, and semantic conditioning.", + "extras": "sampler: Default, cfg_scale: 4.0, steps: 50", + "size": 19.11, + "date": "2026 January" }, "ZAI GLM-Image": { "path": "zai-org/GLM-Image", @@ -889,14 +796,6 @@ "size": 10.46, "date": "2026 January" }, - "Bria FIBO": { - "path": "briaai/FIBO", - "preview": "briaai--FIBO.jpg", - "desc": "BRIA FIBO is an 8-billion parameter text-to-image diffusion model using Flow Matching and featuring a lightweight SmolLM3-3B text encoder. Delivers high-quality, detailed image generation with efficient inference.", - "extras": "sampler: Default, cfg_scale: 3.5", - "size": 23.79, - "date": "2025 October" - }, "Bria Fibo-Edit": { "path": "briaai/Fibo-Edit", "preview": "briaai--Fibo-Edit.jpg", @@ -905,20 +804,59 @@ "size": 22.47, "date": "2026 January" }, - "Bria 3.2": { - "path": "SahilCarterr/BRIA-3.2", - "desc": "Bria 3.2 is the next-generation commercial-ready text-to-image model. With just 4 billion parameters, it provides exceptional aesthetics and text rendering, evaluated to provide on par results to leading open-source models, and outperforming other licensed models.", - "preview": "briaai--BRIA-3.2.jpg", - "size": 18.11, - "date": "2025 July" + "NucleusAI Nucleus-Image": { + "path": "NucleusAI/Nucleus-Image", + "preview": "NucleusAI--Nucleus-Image.jpg", + "desc": "Nucleus-Image is a sparse mixture-of-experts diffusion transformer with 17B total parameters and about 2B active parameters per forward pass, using a Qwen3-VL text encoder and Qwen-Image VAE for 1024px generation.", + "variant": "bf16", + "extras": "sampler: Default, cfg_scale: 8.0, steps: 50", + "size": 48.09, + "date": "2026 March" }, - "StepFun Step1X-Edit v1.1": { - "path": "stepfun-ai/Step1X-Edit-v1p1-diffusers", - "preview": "stepfun-ai--Step1X-Edit-v1p1-diffusers.jpg", - "desc": "Multimodal image editing model using Step1X transformer architecture with Qwen2.5-VL text encoding, trained with Flow Matching scheduler for high-quality in-context image edits and refinements.", + "Baidu ERNIE-Image": { + "path": "baidu/ERNIE-Image", + "preview": "baidu--ERNIE-Image.jpg", + "desc": "ERNIE-Image is a text-to-image diffusion transformer model that combines a Mistral3 text encoder with a FlowMatch transformer and Flux2-style VAE for 1024px image generation.", + "extras": "sampler: Default, cfg_scale: 4.0, steps: 50", + "size": 22.29, + "date": "2026 April" + }, + "JoyAI Image Edit": { + "path": "jdopensource/JoyAI-Image-Edit-Diffusers", + "preview": "jdopensource--JoyAI-Image-Edit-Diffusers.jpg", + "desc": "JoyAI Image Edit is a Diffusers-native image editing model that combines a JoyImageEdit transformer with Qwen3-VL multimodal conditioning for instruction-guided edits.", + "size": 46.86, "extras": "sampler: Default", - "size": 38.91, - "date": "2025 August" + "date": "2026 April" + }, + "Microsoft Lens": { + "path": "Jinstudio/Lens", + "preview": "microsoft--Lens.jpg", + "desc": "Microsoft Lens is a text-to-image DiT model using GPT-OSS chat-style prompt encoding and Flux2 VAE decoding.", + "size": 28.43, + "date": "2026 May" + }, + "Microsoft Lens Base": { + "path": "Jinstudio/Lens-Base", + "preview": "microsoft--Lens-Base.jpg", + "desc": "Microsoft Lens-Base is the base variant of Lens for text-to-image generation with GPT-OSS prompt features.", + "size": 28.43, + "date": "2026 May" + }, + "Anima 1.0 Base": { + "path": "vladmandic/Anima-1.0-Base", + "preview": "vladmandic--Anima-1.0-Base.jpg", + "desc": "Anima 1.0 Base with extended 1024-resolution training and expanded dataset coverage for less common artists. A 2B parameter anime-focused text-to-image model based on modified Cosmos-Predict-2B with Qwen3-0.6B text encoder, created by CircleStone Labs and Comfy Org.", + "date": "2026 May", + "size": 4.99 + }, + "HiDream-O1 Image": { + "path": "HiDream-ai/HiDream-O1-Image", + "desc": "HiDream-O1-Image is an 8B pixel-level unified transformer model for text-to-image generation, instruction editing, and multi-reference personalization up to 2048x2048.", + "preview": "HiDream-ai--HiDream-O1-Image.jpg", + "extras": "sampler: Default", + "size": 35.2, + "date": "2026 May" }, "VIBE Image Edit": { "path": "vladmandic/VIBE-Image-Edit", @@ -928,13 +866,50 @@ "size": 8.63, "date": "2026 May" }, - "JoyAI Image Edit": { - "path": "jdopensource/JoyAI-Image-Edit-Diffusers", - "preview": "jdopensource--JoyAI-Image-Edit-Diffusers.jpg", - "desc": "JoyAI Image Edit is a Diffusers-native image editing model that combines a JoyImageEdit transformer with Qwen3-VL multimodal conditioning for instruction-guided edits.", - "size": 46.86, - "extras": "sampler: Default", - "date": "2026 April" + "Boogu Image 0.1 Base": { + "path": "Boogu/Boogu-Image-0.1-Base", + "preview": "Boogu--Boogu-Image-0.1-Base.jpg", + "desc": "Boogu Image 0.1 Base is the core Boogu text-to-image model with Qwen3-VL instruction encoding and a mixed single/double-stream diffusion transformer.", + "size": 35.81, + "date": "2026 June" + }, + "Boogu Image 0.1 Edit": { + "path": "Boogu/Boogu-Image-0.1-Edit", + "preview": "Boogu--Boogu-Image-0.1-Edit.jpg", + "desc": "Boogu Image 0.1 Edit is optimized for image editing with the same instruction-aware Boogu transformer and flow-match scheduler.", + "size": 35.81, + "date": "2026 June" + }, + "Ideogram 4": { + "path": "CalamitousFelicitousness/Ideogram-4-bf16-Diffusers", + "preview": "CalamitousFelicitousness--Ideogram-4-bf16-Diffusers.jpg", + "desc": "Ideogram 4 is Ideogram's first open-weight text-to-image model: a two 9.3B flow-matching DiTs that uses a Qwen3-VL vision-language model as its text encoder, with strong in-image text rendering. Requires structured JSON-caption prompts; prompt-enhance (on by default) rewrites a plain prompt into one.", + "extras": "sampler: Default, cfg_scale: 7.0, steps: 20, width: 1024, height: 1024", + "size": 49.9, + "date": "2026 June" + }, + "Krea 2 Base": { + "path": "CalamitousFelicitousness/Krea-2-Base-Diffusers", + "preview": "CalamitousFelicitousness--Krea-2-Base-Diffusers.jpg", + "desc": "Krea 2 (K2) Base is the undistilled foundation model of the Krea 2 family, trained from scratch by Krea. A 12.9B-parameter single-stream flow-matching DiT that uses a Qwen3-VL-4B vision-language model as its text encoder and the Qwen-Image VAE. The base checkpoint is intended for fine-tuning and LoRA training; LoRAs trained on it apply to Krea 2 Turbo.", + "extras": "sampler: Default, cfg_scale: 4.5, steps: 52", + "size": 33.5, + "date": "2026 June" + }, + "Photoroom PRXPixel": { + "path": "Photoroom/prxpixel-t2i", + "desc": "PRXPixel is a 7-billion-parameter pixel-space text-to-image model that denoises RGB images directly using a Qwen3-VL text encoder and flow-matching scheduler.", + "preview": "Photoroom--prxpixel-t2i.jpg", + "size": 16.25, + "date": "2026 June" + }, + "Anima 1.0 Aesthetic": { + "path": "CalamitousFelicitousness/Anima-1.0-Aesthetic-Diffusers", + "preview": "CalamitousFelicitousness--Anima-1.0-Aesthetic-Diffusers.jpg", + "desc": "Anima 1.0 Aesthetic, fine-tuned for better consistency and a higher quality default art style. A 2B parameter anime-focused text-to-image model based on modified Cosmos-Predict-2B with Qwen3-0.6B text encoder, created by CircleStone Labs and Comfy Org.", + "extras": "sampler: Default, cfg_scale: 4.5, steps: 30", + "date": "2026 July", + "size": 4.99 }, "JoyAI Image Edit Plus": { "path": "jdopensource/JoyAI-Image-Edit-Plus-Diffusers", @@ -983,5 +958,37 @@ "extras": "sampler: Default", "size": 17.69, "date": "2026 July" + }, + "Anima 2.9B Preview v1": { + "path": "CalamitousFelicitousness/Anima-2.9B-Preview-v1-Diffusers", + "preview": "CalamitousFelicitousness--Anima-2.9B-Preview-v1-Diffusers.jpg", + "desc": "Anima 2.9B preview v1 by Gazingstars, a depth-expanded fine-tune of Anima 1.0 Base: the transformer is grown from 28 to 40 layers and trained on 1.7M additional anime and illustration samples with mixed tag and natural-language captions. Same Qwen3-0.6B text encoder, LLM adapter and VAE as Anima 1.0.", + "extras": "sampler: Default, cfg_scale: 4.0, steps: 30", + "date": "2026 August", + "size": 7.32 + }, + "MiniMaxAI MiniMax-H3": { + "path": "MiniMaxAI/MiniMax-H3", + "preview": "MiniMaxAI--MiniMax-H3.jpg", + "desc": "MiniMax-H3 generates video with synchronized stereo audio in a single denoising pass through a 33B single-stream transformer with a Qwen3-VL conditioner. In image tabs the model runs in experimental still mode, keeping the first frame of a minimal generation.", + "extras": "sampler: Default", + "size": 134.12, + "date": "2026 August" + }, + "MiniMaxAI MiniMax-H3 Ref2VA": { + "path": "MiniMaxAI/MiniMax-H3", + "subfolder": "ref2va", + "preview": "MiniMaxAI--MiniMax-H3.jpg", + "desc": "The omni-reference variant of MiniMax-H3, sharing one repository with the base model as a separate checkpoint partition. Video with synchronized stereo audio is conditioned on reference images for identity and appearance, with reference rows held clean while video rows denoise.", + "extras": "sampler: Default", + "size": 134.12, + "date": "2026 August" + }, + "inclusionAI LLaDA-Image": { + "path": "inclusionAI/LLaDA-Image", + "desc": "LLaDA-Image is a unified 6B image generation and editing model with text, VQ-conditioned, and reference-image workflows.", + "extras": "steps: 50, cfg_scale: 5.0", + "size": 37.15, + "date": "2026 September" } } diff --git a/data/reference-cloud.json b/data/reference-cloud.json index d08526559..10f0e5930 100644 --- a/data/reference-cloud.json +++ b/data/reference-cloud.json @@ -1,22 +1,22 @@ { - "Nano Banana": { - "path": "gemini-2.5-flash-image", - "desc": "Our best engine for high-velocity visual creation, offering state-of-the-art speed and efficiency. Gemini 2.5 Flash Image, also known as Nano Banana, is best for high-volume generation, conversational image editing, and low-latency creative workflows that require native multimodal understanding. (Knowledge cutoff June 2025)", - "preview": "gemini-2.5-flash-image.jpg" + "Nano Banana lite": { + "path": "gemini-3.1-flash-lite-image", + "desc": "Nano Banana Lite is designed as the efficiency specialist of the image generation family, offering ultra-low latency and cost-effective image generation and editing. By targeting a sub-2 second latency and significantly reduced TPU compute costs, this model enables high-volume interactive developer use cases and real-time consumer applications. (Knowledge cutoff January 2025)", + "preview": "gemini-3.1-flash-lite-image.jpg" }, "Nano Banana 2": { "path": "gemini-3.1-flash-image", "desc": "Nano Banana 2 provides high-quality image generation and conversational editing at a mainstream price point and low latency. It serves as the high-efficiency counterpart to Gemini 3 Pro Image, optimized for speed and high-volume developer use cases.(Knowledge cutoff January 2025)", "preview": "gemini-3.1-flash-image.jpg" }, - "Nano Banana lite": { - "path": "gemini-3.1-flash-lite-image", - "desc": "Nano Banana Lite is designed as the efficiency specialist of the image generation family, offering ultra-low latency and cost-effective image generation and editing. By targeting a sub-2 second latency and significantly reduced TPU compute costs, this model enables high-volume interactive developer use cases and real-time consumer applications. (Knowledge cutoff January 2025)", - "preview": "gemini-3.1-flash-lite-image.jpg" - }, "Nano Banana Pro": { "path": "gemini-3-pro-image", "desc": "Nano Banana Pro is a sophisticated reasoning-driven engine for professional-grade image editing and generation, offering studio-quality precision and advanced creative control. Nano Banana Pro is best for complex graphic design, high-fidelity product mockups, and factual data visualizations that require accurate text rendering and real-world grounding via Google Search. (Knowledge cutoff January 2025)", "preview": "gemini-3-pro-image.jpg" + }, + "Nano Banana": { + "path": "gemini-2.5-flash-image", + "desc": "Our best engine for high-velocity visual creation, offering state-of-the-art speed and efficiency. Gemini 2.5 Flash Image, also known as Nano Banana, is best for high-volume generation, conversational image editing, and low-latency creative workflows that require native multimodal understanding. (Knowledge cutoff June 2025)", + "preview": "gemini-2.5-flash-image.jpg" } } diff --git a/data/reference-community.json b/data/reference-community.json index 2fc9d5290..6fa5c68c7 100644 --- a/data/reference-community.json +++ b/data/reference-community.json @@ -1,17 +1,12 @@ { - "Tempest-by-Vlad XL": { - "path": "tempestByVlad_baseV01.safetensors@https://civitai.com/api/download/models/1301775", - "preview": "tempestByVlad_baseV01.jpg", - "desc": "Flexible SDXL model with custom encoder and finetuned for larger landscape resolutions with high details and high contrast.", - "size": 6.94, - "date": "2025 January" - }, - "Tempest-by-Vlad XL Hyper": { - "path": "tempestByVlad_hyperV01.safetensors@https://civitai.com/api/download/models/1343512", - "preview": "tempestByVlad_hyperV01.jpg", - "desc": "Custom distilled variant with goal to get as-normal-as-possible model that works with low steps and guidance-free", - "size": 6.94, - "date": "2025 January" + "Juggernaut SD Reborn": { + "original": true, + "path": "juggernaut_reborn.safetensors@https://civitai.com/api/download/models/274039", + "preview": "juggernaut_reborn.jpg", + "desc": "Showcase finetuned model based on Stable diffusion 1.5", + "date": "2023 December", + "size": 2.28, + "extras": "width: 512, height: 512, sampler: DEIS, steps: 20, cfg_scale: 6.0" }, "Juggernaut XL XI": { "path": "juggernautXL_juggXIByRundiffusion.safetensors@https://civitai.com/api/download/models/782002", @@ -29,28 +24,26 @@ "size": 6.94, "extras": "sampler: DPM SDE, steps: 6, cfg_scale: 2.0" }, - "Juggernaut SD Reborn": { - "original": true, - "path": "juggernaut_reborn.safetensors@https://civitai.com/api/download/models/274039", - "preview": "juggernaut_reborn.jpg", - "desc": "Showcase finetuned model based on Stable diffusion 1.5", - "date": "2023 December", - "size": 2.28, - "extras": "width: 512, height: 512, sampler: DEIS, steps: 20, cfg_scale: 6.0" - }, - "WAI Illustrious XL v15": { - "path": "waiIllustriousSDXL_v150.safetensors@https://civitai.com/api/download/models/2167369", - "preview": "waiIllustriousSDXL_v150.jpg", + "NoobAI XL 1.1 Epsilon": { + "path": "noobaiXLNAIXL_epsilonPred11Version.safetensors@https://huggingface.co/Laxhar/noobai-XL-1.1/resolve/main/NoobAI-XL-v1.1.safetensors", + "preview": "noobaiXLNAIXL_epsilonPred11Version.jpg", "desc": "", "size": 6.94, - "date": "2025 August" + "date": "2024 November" }, - "Pony Realism XL v2.3": { - "path": "ponyRealism_V23.safetensors@https://civitai.com/api/download/models/1763661", - "preview": "ponyRealism_V23.jpg", - "desc": "", - "size": 6.94, - "date": "2025 May" + "ShuttleAI Shuttle 3.0 Diffusion": { + "path": "shuttleai/shuttle-3-diffusion", + "desc": "Shuttle uses Flux.1 Schnell as its base. It can produce images similar to Flux Dev or Pro in just 4 steps, and it is licensed under Apache 2. The model was partially de-distilled during training. When used beyond 10 steps, it enters refiner mode enhancing image details without altering the composition", + "preview": "shuttleai--shuttle-3-diffusion.jpg", + "date": "2024 November", + "size": 31.41 + }, + "ShuttleAI Shuttle 3.1 Aesthetic": { + "path": "shuttleai/shuttle-3.1-aesthetic", + "desc": "Shuttle uses Flux.1 Schnell as its base. It can produce images similar to Flux Dev or Pro in just 4 steps, and it is licensed under Apache 2. The model was partially de-distilled during training. When used beyond 10 steps, it enters refiner mode enhancing image details without altering the composition", + "preview": "shuttleai--shuttle-3.1-aesthetic.jpg", + "date": "2024 November", + "size": 31.41 }, "NoobAI XL 1.0 V-Pred": { "path": "noobaiXLNAIXL_vPred10Version.safetensors@https://huggingface.co/Laxhar/noobai-XL-Vpred-1.0/resolve/main/NoobAI-XL-Vpred-v1.0.safetensors", @@ -59,12 +52,33 @@ "size": 6.94, "date": "2024 December" }, - "NoobAI XL 1.1 Epsilon": { - "path": "noobaiXLNAIXL_epsilonPred11Version.safetensors@https://huggingface.co/Laxhar/noobai-XL-1.1/resolve/main/NoobAI-XL-v1.1.safetensors", - "preview": "noobaiXLNAIXL_epsilonPred11Version.jpg", + "Tempest-by-Vlad XL": { + "path": "tempestByVlad_baseV01.safetensors@https://civitai.com/api/download/models/1301775", + "preview": "tempestByVlad_baseV01.jpg", + "desc": "Flexible SDXL model with custom encoder and finetuned for larger landscape resolutions with high details and high contrast.", + "size": 6.94, + "date": "2025 January" + }, + "Tempest-by-Vlad XL Hyper": { + "path": "tempestByVlad_hyperV01.safetensors@https://civitai.com/api/download/models/1343512", + "preview": "tempestByVlad_hyperV01.jpg", + "desc": "Custom distilled variant with goal to get as-normal-as-possible model that works with low steps and guidance-free", + "size": 6.94, + "date": "2025 January" + }, + "ShuttleAI Shuttle Jaguar": { + "path": "shuttleai/shuttle-jaguar", + "desc": "Shuttle uses Flux.1 Schnell as its base. It can produce images similar to Flux Dev or Pro in just 4 steps, and it is licensed under Apache 2. The model was partially de-distilled during training. When used beyond 10 steps, it enters refiner mode enhancing image details without altering the composition", + "preview": "shuttleai--shuttle-jaguar.jpg", + "date": "2025 January", + "size": 31.41 + }, + "Pony Realism XL v2.3": { + "path": "ponyRealism_V23.safetensors@https://civitai.com/api/download/models/1763661", + "preview": "ponyRealism_V23.jpg", "desc": "", "size": 6.94, - "date": "2024 November" + "date": "2025 May" }, "WAI-Ani-Pony XL v14": { "path": "waiANIPONYXL_v140.safetensors@https://civitai.com/api/download/models/1767402", @@ -73,6 +87,55 @@ "size": 6.94, "date": "2025 May" }, + "WAI Illustrious XL v15": { + "path": "waiIllustriousSDXL_v150.safetensors@https://civitai.com/api/download/models/2167369", + "preview": "waiIllustriousSDXL_v150.jpg", + "desc": "", + "size": 6.94, + "date": "2025 August" + }, + "Tiwaz CenKreChro": { + "path": "Tiwaz/CenKreChro", + "preview": "Tiwaz--CenKreChro.jpg", + "desc": "Based Centerfold Flux 5, trying to merge in Chroma and Krea.", + "date": "2025 September", + "size": 31.42 + }, + "purplesmartai Pony 7": { + "path": "purplesmartai/pony-v7-base", + "preview": "purplesmartai--pony-v7-base.jpg", + "desc": "Pony V7 is a versatile character generation model based on AuraFlow architecture. It supports a wide range of styles and species types (humanoid, anthro, feral, and more) and handles character interactions through natural language prompts.", + "date": "2025 October", + "size": 33.32 + }, + "Skywork UniPic3": { + "path": "Skywork/Unipic3", + "preview": "Skywork--Unipic3.jpg", + "desc": "UniPic3 is an image editing and multi-image composition model based. It is a fine-tune of Qwen-Image-Edit.", + "date": "2026 January", + "size": 53.74 + }, + "Skywork Unipic3-DMD": { + "path": "Skywork/Unipic3-DMD", + "preview": "Skywork--Unipic3-DMD.jpg", + "desc": "UniPic3-DMD-Model is a few-step image editing and multi-image composition model trained using Distribution Matching Distillation (DMD) and is a fine-tune of Qwen-Image-Edit.", + "date": "2026 January", + "size": 53.74 + }, + "FireRed Image Edit 1.0": { + "path": "FireRedTeam/FireRed-Image-Edit-1.0", + "preview": "FireRedTeam--FireRed-Image-Edit-1.0.jpg", + "desc": "FireRed-Image-Edit is a general-purpose image editing model that delivers high-fidelity and consistent editing across a wide range of scenarios. FireRed is a fine-tune of Qwen-Image-Edit.", + "date": "2026 February", + "size": 53.74 + }, + "FireRed Image Edit 1.1": { + "path": "FireRedTeam/FireRed-Image-Edit-1.1", + "preview": "FireRedTeam--FireRed-Image-Edit-1.1.jpg", + "desc": "FireRed-Image-Edit is a general-purpose image editing model that delivers high-fidelity and consistent editing across a wide range of scenarios. FireRed is a fine-tune of Qwen-Image-Edit.", + "date": "2026 March", + "size": 53.74 + }, "Z-Image-Turbo MoodyRealMix": { "path": "resonantsky/MoodyRealMix-SDNQ-int8-svd-r32", "preview": "resonantsky--MoodyRealMix-SDNQ-int8-svd-r32.jpg", @@ -100,69 +163,6 @@ "tags": "community, Z-image", "date": "2026 May" }, - "Tiwaz CenKreChro": { - "path": "Tiwaz/CenKreChro", - "preview": "Tiwaz--CenKreChro.jpg", - "desc": "Based Centerfold Flux 5, trying to merge in Chroma and Krea.", - "date": "2025 September", - "size": 31.42 - }, - "purplesmartai Pony 7": { - "path": "purplesmartai/pony-v7-base", - "preview": "purplesmartai--pony-v7-base.jpg", - "desc": "Pony V7 is a versatile character generation model based on AuraFlow architecture. It supports a wide range of styles and species types (humanoid, anthro, feral, and more) and handles character interactions through natural language prompts.", - "date": "2025 October", - "size": 33.32 - }, - "ShuttleAI Shuttle 3.0 Diffusion": { - "path": "shuttleai/shuttle-3-diffusion", - "desc": "Shuttle uses Flux.1 Schnell as its base. It can produce images similar to Flux Dev or Pro in just 4 steps, and it is licensed under Apache 2. The model was partially de-distilled during training. When used beyond 10 steps, it enters refiner mode enhancing image details without altering the composition", - "preview": "shuttleai--shuttle-3-diffusion.jpg", - "date": "2024 November", - "size": 31.41 - }, - "ShuttleAI Shuttle 3.1 Aesthetic": { - "path": "shuttleai/shuttle-3.1-aesthetic", - "desc": "Shuttle uses Flux.1 Schnell as its base. It can produce images similar to Flux Dev or Pro in just 4 steps, and it is licensed under Apache 2. The model was partially de-distilled during training. When used beyond 10 steps, it enters refiner mode enhancing image details without altering the composition", - "preview": "shuttleai--shuttle-3.1-aesthetic.jpg", - "date": "2024 November", - "size": 31.41 - }, - "ShuttleAI Shuttle Jaguar": { - "path": "shuttleai/shuttle-jaguar", - "desc": "Shuttle uses Flux.1 Schnell as its base. It can produce images similar to Flux Dev or Pro in just 4 steps, and it is licensed under Apache 2. The model was partially de-distilled during training. When used beyond 10 steps, it enters refiner mode enhancing image details without altering the composition", - "preview": "shuttleai--shuttle-jaguar.jpg", - "date": "2025 January", - "size": 31.41 - }, - "FireRed Image Edit 1.0": { - "path": "FireRedTeam/FireRed-Image-Edit-1.0", - "preview": "FireRedTeam--FireRed-Image-Edit-1.0.jpg", - "desc": "FireRed-Image-Edit is a general-purpose image editing model that delivers high-fidelity and consistent editing across a wide range of scenarios. FireRed is a fine-tune of Qwen-Image-Edit.", - "date": "2026 February", - "size": 53.74 - }, - "FireRed Image Edit 1.1": { - "path": "FireRedTeam/FireRed-Image-Edit-1.1", - "preview": "FireRedTeam--FireRed-Image-Edit-1.1.jpg", - "desc": "FireRed-Image-Edit is a general-purpose image editing model that delivers high-fidelity and consistent editing across a wide range of scenarios. FireRed is a fine-tune of Qwen-Image-Edit.", - "date": "2026 March", - "size": 53.74 - }, - "Skywork UniPic3": { - "path": "Skywork/Unipic3", - "preview": "Skywork--Unipic3.jpg", - "desc": "UniPic3 is an image editing and multi-image composition model based. It is a fine-tune of Qwen-Image-Edit.", - "date": "2026 January", - "size": 53.74 - }, - "Skywork Unipic3-DMD": { - "path": "Skywork/Unipic3-DMD", - "preview": "Skywork--Unipic3-DMD.jpg", - "desc": "UniPic3-DMD-Model is a few-step image editing and multi-image composition model trained using Distribution Matching Distillation (DMD) and is a fine-tune of Qwen-Image-Edit.", - "date": "2026 January", - "size": 53.74 - }, "Anima 1.0 Base Merge sdnq-hadamard-uint4": { "path": "vladmandic/Anima-1.0-Base-Merge-sdnq-hadamard-uint4", "preview": "vladmandic--Anima-1.0-Base.jpg", diff --git a/data/reference-distilled.json b/data/reference-distilled.json index 96f70d453..b1f07f95e 100644 --- a/data/reference-distilled.json +++ b/data/reference-distilled.json @@ -1,17 +1,20 @@ { - "Boogu Image 0.1 Turbo": { - "path": "Boogu/Boogu-Image-0.1-Turbo", - "preview": "Boogu--Boogu-Image-0.1-Turbo.jpg", - "desc": "Boogu Image 0.1 Turbo is the distilled fast inference variant of Boogu Image with the same Qwen3-VL instruction encoder and Boogu transformer architecture.", - "size": 35.81, - "date": "2026 June" + "Segmind Tiny": { + "path": "segmind/tiny-sd", + "preview": "segmind--tiny-sd.jpg", + "desc": "Segmind's Tiny-SD offers a compact, efficient, and distilled version of Realistic Vision 4.0 and is up to 80% faster than SD1.5", + "extras": "width: 512, height: 512, sampler: Default, cfg_scale: 9.0", + "size": 0.99, + "date": "2023 July" }, - "Boogu Image 0.1 Edit Turbo": { - "path": "Boogu/Boogu-Image-0.1-Edit-Turbo", - "preview": "Boogu--Boogu-Image-0.1-Edit-Turbo.jpg", - "desc": "Boogu Image 0.1 Edit Turbo is the distilled editing variant of Boogu Image with motion-aware instruction encoding and fast flow-match inference.", - "size": 35.81, - "date": "2026 June" + "Segmind SSD-1B": { + "path": "huggingface/segmind/SSD-1B", + "preview": "segmind--SSD-1B.jpg", + "desc": "The Segmind Stable Diffusion Model (SSD-1B) offers a compact, efficient, and distilled version of the SDXL model. At 50% smaller and 60% faster than Stable Diffusion XL (SDXL), it provides quick and seamless performance without sacrificing image quality.", + "variant": "fp16", + "extras": "sampler: Default, cfg_scale: 9.0", + "size": 12.48, + "date": "2023 October" }, "StabilityAI StableDiffusion XL Turbo": { "path": "stabilityai/sdxl-turbo", @@ -22,13 +25,21 @@ "size": 19.38, "date": "2023 November" }, - "Krea 2 Turbo": { - "path": "CalamitousFelicitousness/Krea-2-Turbo-Diffusers", - "preview": "CalamitousFelicitousness--Krea-2-Turbo-Diffusers.jpg", - "desc": "Krea 2 (K2) Turbo is the 8-step distilled inference model of the Krea 2 family, trained from scratch by Krea. A 12.9B-parameter single-stream flow-matching DiT that uses a Qwen3-VL-4B vision-language model as its text encoder and the Qwen-Image VAE. Runs without classifier-free guidance; LoRAs trained on Krea 2 Base apply directly.", - "extras": "sampler: Default, cfg_scale: 1.0, steps: 8", - "size": 33.5, - "date": "2026 June" + "Tencent HunyuanDiT 1.1 Distilled": { + "path": "Tencent-Hunyuan/HunyuanDiT-v1.1-Diffusers-Distilled", + "desc": "Hunyuan-DiT : A Powerful Multi-Resolution Diffusion Transformer with Fine-Grained Chinese Understanding.", + "preview": "Tencent-Hunyuan--HunyuanDiT-v1.1-Diffusers-Distilled.jpg", + "extras": "sampler: Default, cfg_scale: 2.0", + "size": 13.49, + "date": "2024 June" + }, + "Tencent HunyuanDiT 1.2 Distilled": { + "path": "Tencent-Hunyuan/HunyuanDiT-v1.2-Diffusers-Distilled", + "desc": "Hunyuan-DiT : A Powerful Multi-Resolution Diffusion Transformer with Fine-Grained Chinese Understanding.", + "preview": "Tencent-Hunyuan--HunyuanDiT-v1.2-Diffusers-Distilled.jpg", + "extras": "sampler: Default, cfg_scale: 2.0", + "size": 13.43, + "date": "2024 July" }, "StabilityAI Stable Diffusion 3.5 Turbo": { "path": "stabilityai/stable-diffusion-3.5-large-turbo", @@ -39,28 +50,12 @@ "size": 36.12, "date": "2024 October" }, - "Microsoft Lens Turbo": { - "path": "Jinstudio/Lens-Turbo", - "preview": "microsoft--Lens-Turbo.jpg", - "desc": "Microsoft Lens-Turbo is the distilled Lens variant optimized for faster text-to-image generation with fewer steps.", - "size": 28.43, - "date": "2026 May" - }, - "Tencent FLUX.1 Dev SRPO": { - "path": "vladmandic/flux.1-dev-SRPO", - "preview": "vladmandic--flux.1-dev-SRPO.jpg", - "desc": "FLUX.1 Dev SRPO is Tencent trained with specific technique: Directly Aligning the Full Diffusion Trajectory with Fine-Grained Human Preference", - "extras": "sampler: Default, cfg_scale: 4.5", - "size": 31.42, - "date": "2025 September" - }, - "HiDream-O1 Image Dev": { - "path": "HiDream-ai/HiDream-O1-Image-Dev", - "preview": "HiDream-ai--HiDream-O1-Image-Dev.jpg", - "desc": "HiDream-O1-Image-Dev is the distilled 8B HiDream-O1 variant tuned for 28-step fast generation using flash flow scheduling.", - "extras": "sampler: Flash, steps: 28, cfg_scale: 0.0", - "size": 35.2, - "date": "2026 May" + "NVLabs Sana 1.5 1.6B 1k Sprint": { + "path": "Efficient-Large-Model/Sana_Sprint_1.6B_1024px_diffusers", + "desc": "SANA-Sprint is an ultra-efficient diffusion model for text-to-image (T2I) generation, reducing inference steps from 20 to 1-4 while achieving state-of-the-art performance.", + "preview": "Efficient-Large-Model--Sana15_Sprint_1600M_1024px_diffusers.jpg", + "size": 9.03, + "date": "2025 March" }, "Qwen-Image-Lightning": { "path": "vladmandic/Qwen-Lightning", @@ -78,13 +73,20 @@ "size": 56.1, "date": "2025 August" }, - "Baidu ERNIE-Image-Turbo": { - "path": "baidu/ERNIE-Image-Turbo", - "preview": "baidu--ERNIE-Image-Turbo.jpg", - "desc": "ERNIE-Image-Turbo is a distilled ERNIE-Image variant optimized for fast generation with fewer denoising steps.", - "extras": "sampler: Default, cfg_scale: 1.0, steps: 8", - "size": 22.29, - "date": "2026 April" + "lodestones Chroma1 Flash": { + "path": "lodestones/Chroma1-Flash", + "preview": "lodestones--Chroma1-Flash.jpg", + "desc": "Chroma is a 8.9B parameter model based on FLUX.1-schnell. It’s fully Apache 2.0 licensed, ensuring that anyone can use, modify, and build on top of it—no corporate gatekeeping. A fine-tuned version of the Chroma1-Base made to find the best way to make these flow matching models faster.", + "size": 25.6, + "date": "2025 August" + }, + "Tencent FLUX.1 Dev SRPO": { + "path": "vladmandic/flux.1-dev-SRPO", + "preview": "vladmandic--flux.1-dev-SRPO.jpg", + "desc": "FLUX.1 Dev SRPO is Tencent trained with specific technique: Directly Aligning the Full Diffusion Trajectory with Fine-Grained Human Preference", + "extras": "sampler: Default, cfg_scale: 4.5", + "size": 31.42, + "date": "2025 September" }, "Qwen-Image-Lightning-Edit": { "path": "vladmandic/Qwen-Lightning-Edit", @@ -110,6 +112,13 @@ "date": "2025 September", "size": 41.08 }, + "Tencent HunyuanImage 2.1 Distilled": { + "path": "hunyuanvideo-community/HunyuanImage-2.1-Distilled-Diffusers", + "desc": "HunyuanImage-2.1, a highly efficient text-to-image model that is capable of generating 2K (2048 × 2048) resolution images.", + "preview": "hunyuanvideo-community--HunyuanImage-2.1-Distilled-Diffusers.jpg", + "size": 49.53, + "date": "2025 September" + }, "Qwen-Image-Edit-2509 Pruning-13B": { "path": "OPPOer/Qwen-Image-Edit-2509-Pruning", "subfolder": "Qwen-Image-Edit-2509-13B-4steps", @@ -118,51 +127,6 @@ "date": "2025 October", "size": 42.34 }, - "lodestones Chroma1 Flash": { - "path": "lodestones/Chroma1-Flash", - "preview": "lodestones--Chroma1-Flash.jpg", - "desc": "Chroma is a 8.9B parameter model based on FLUX.1-schnell. It’s fully Apache 2.0 licensed, ensuring that anyone can use, modify, and build on top of it—no corporate gatekeeping. A fine-tuned version of the Chroma1-Base made to find the best way to make these flow matching models faster.", - "size": 25.6, - "date": "2025 August" - }, - "SDXL Flash Mini": { - "path": "SDXL-Flash_Mini.safetensors@https://huggingface.co/sd-community/sdxl-flash-mini/resolve/main/SDXL-Flash_Mini.safetensors?download=true", - "preview": "SDXL-Flash_Mini.jpg", - "desc": "Introducing the new fast model SDXL Flash (Mini), we learned that all fast XL models work fast, but the quality decreases, and we also made a fast model, but it is not as fast as LCM, Turbo, Lightning and Hyper, but the quality is higher.", - "extras": "sampler: DEIS, steps: 40, cfg_scale: 6.0", - "experimental": true - }, - "NVLabs Sana 1.5 1.6B 1k Sprint": { - "path": "Efficient-Large-Model/Sana_Sprint_1.6B_1024px_diffusers", - "desc": "SANA-Sprint is an ultra-efficient diffusion model for text-to-image (T2I) generation, reducing inference steps from 20 to 1-4 while achieving state-of-the-art performance.", - "preview": "Efficient-Large-Model--Sana15_Sprint_1600M_1024px_diffusers.jpg", - "size": 9.03, - "date": "2025 March" - }, - "Segmind SSD-1B": { - "path": "huggingface/segmind/SSD-1B", - "preview": "segmind--SSD-1B.jpg", - "desc": "The Segmind Stable Diffusion Model (SSD-1B) offers a compact, efficient, and distilled version of the SDXL model. At 50% smaller and 60% faster than Stable Diffusion XL (SDXL), it provides quick and seamless performance without sacrificing image quality.", - "variant": "fp16", - "extras": "sampler: Default, cfg_scale: 9.0", - "size": 12.48, - "date": "2023 October" - }, - "Segmind Tiny": { - "path": "segmind/tiny-sd", - "preview": "segmind--tiny-sd.jpg", - "desc": "Segmind's Tiny-SD offers a compact, efficient, and distilled version of Realistic Vision 4.0 and is up to 80% faster than SD1.5", - "extras": "width: 512, height: 512, sampler: Default, cfg_scale: 9.0", - "size": 0.99, - "date": "2023 July" - }, - "Tencent HunyuanImage 2.1 Distilled": { - "path": "hunyuanvideo-community/HunyuanImage-2.1-Distilled-Diffusers", - "desc": "HunyuanImage-2.1, a highly efficient text-to-image model that is capable of generating 2K (2048 × 2048) resolution images.", - "preview": "hunyuanvideo-community--HunyuanImage-2.1-Distilled-Diffusers.jpg", - "size": 49.53, - "date": "2025 September" - }, "Bria Fibo-Lite": { "path": "briaai/Fibo-lite", "preview": "briaai--Fibo-lite.jpg", @@ -171,22 +135,6 @@ "size": 22.47, "date": "2025 November" }, - "Tencent HunyuanDiT 1.2 Distilled": { - "path": "Tencent-Hunyuan/HunyuanDiT-v1.2-Diffusers-Distilled", - "desc": "Hunyuan-DiT : A Powerful Multi-Resolution Diffusion Transformer with Fine-Grained Chinese Understanding.", - "preview": "Tencent-Hunyuan--HunyuanDiT-v1.2-Diffusers-Distilled.jpg", - "extras": "sampler: Default, cfg_scale: 2.0", - "size": 13.43, - "date": "2024 July" - }, - "Tencent HunyuanDiT 1.1 Distilled": { - "path": "Tencent-Hunyuan/HunyuanDiT-v1.1-Diffusers-Distilled", - "desc": "Hunyuan-DiT : A Powerful Multi-Resolution Diffusion Transformer with Fine-Grained Chinese Understanding.", - "preview": "Tencent-Hunyuan--HunyuanDiT-v1.1-Diffusers-Distilled.jpg", - "extras": "sampler: Default, cfg_scale: 2.0", - "size": 13.49, - "date": "2024 June" - }, "Black Forest Labs FLUX.2 Klein 4B": { "path": "black-forest-labs/FLUX.2-klein-4B", "preview": "black-forest-labs--FLUX.2-klein-4B.jpg", @@ -203,6 +151,13 @@ "size": 32.32, "date": "2026 January" }, + "Meituan LongCat Image-Edit Turbo": { + "path": "meituan-longcat/LongCat-Image-Edit-Turbo", + "preview": "meituan-longcat--LongCat-Image-Edit.jpg", + "desc": "LongCat-Image-Edit-Turbo, the distilled version of LongCat-Image-Edit. It achieves high-quality image editing with only 8 NFEs (Number of Function Evaluations) , offering extremely low inference latency.", + "size": 27.28, + "date": "2026 February" + }, "Black Forest Labs FLUX.2 Klein 9B KV": { "path": "black-forest-labs/FLUX.2-klein-9b-kv", "preview": "black-forest-labs--FLUX.2-klein-9b-kv.jpg", @@ -211,6 +166,51 @@ "size": 32.32, "date": "2026 March" }, + "Baidu ERNIE-Image-Turbo": { + "path": "baidu/ERNIE-Image-Turbo", + "preview": "baidu--ERNIE-Image-Turbo.jpg", + "desc": "ERNIE-Image-Turbo is a distilled ERNIE-Image variant optimized for fast generation with fewer denoising steps.", + "extras": "sampler: Default, cfg_scale: 1.0, steps: 8", + "size": 22.29, + "date": "2026 April" + }, + "Microsoft Lens Turbo": { + "path": "Jinstudio/Lens-Turbo", + "preview": "microsoft--Lens-Turbo.jpg", + "desc": "Microsoft Lens-Turbo is the distilled Lens variant optimized for faster text-to-image generation with fewer steps.", + "size": 28.43, + "date": "2026 May" + }, + "HiDream-O1 Image Dev": { + "path": "HiDream-ai/HiDream-O1-Image-Dev", + "preview": "HiDream-ai--HiDream-O1-Image-Dev.jpg", + "desc": "HiDream-O1-Image-Dev is the distilled 8B HiDream-O1 variant tuned for 28-step fast generation using flash flow scheduling.", + "extras": "sampler: Flash, steps: 28, cfg_scale: 0.0", + "size": 35.2, + "date": "2026 May" + }, + "Krea 2 Turbo": { + "path": "CalamitousFelicitousness/Krea-2-Turbo-Diffusers", + "preview": "CalamitousFelicitousness--Krea-2-Turbo-Diffusers.jpg", + "desc": "Krea 2 (K2) Turbo is the 8-step distilled inference model of the Krea 2 family, trained from scratch by Krea. A 12.9B-parameter single-stream flow-matching DiT that uses a Qwen3-VL-4B vision-language model as its text encoder and the Qwen-Image VAE. Runs without classifier-free guidance; LoRAs trained on Krea 2 Base apply directly.", + "extras": "sampler: Default, cfg_scale: 1.0, steps: 8", + "size": 33.5, + "date": "2026 June" + }, + "Boogu Image 0.1 Turbo": { + "path": "Boogu/Boogu-Image-0.1-Turbo", + "preview": "Boogu--Boogu-Image-0.1-Turbo.jpg", + "desc": "Boogu Image 0.1 Turbo is the distilled fast inference variant of Boogu Image with the same Qwen3-VL instruction encoder and Boogu transformer architecture.", + "size": 35.81, + "date": "2026 June" + }, + "Boogu Image 0.1 Edit Turbo": { + "path": "Boogu/Boogu-Image-0.1-Edit-Turbo", + "preview": "Boogu--Boogu-Image-0.1-Edit-Turbo.jpg", + "desc": "Boogu Image 0.1 Edit Turbo is the distilled editing variant of Boogu Image with motion-aware instruction encoding and fast flow-match inference.", + "size": 35.81, + "date": "2026 June" + }, "Anima 1.0 Turbo": { "path": "CalamitousFelicitousness/Anima-1.0-Turbo-Diffusers", "preview": "CalamitousFelicitousness--Anima-1.0-Turbo-Diffusers.jpg", @@ -219,13 +219,6 @@ "date": "2026 July", "size": 4.99 }, - "Meituan LongCat Image-Edit Turbo": { - "path": "meituan-longcat/LongCat-Image-Edit-Turbo", - "preview": "meituan-longcat--LongCat-Image-Edit.jpg", - "desc": "LongCat-Image-Edit-Turbo, the distilled version of LongCat-Image-Edit. It achieves high-quality image editing with only 8 NFEs (Number of Function Evaluations) , offering extremely low inference latency.", - "size": 27.28, - "date": "2026 February" - }, "Microsoft Mage-Flow Turbo": { "path": "vladmandic/Mage-Flow-4B-Turbo", "preview": "vladmandic--Mage-Flow-Turbo-4B.jpg", @@ -257,5 +250,19 @@ "extras": "sampler: Default", "size": 17.69, "date": "2026 July" + }, + "inclusionAI LLaDA-Image Turbo": { + "path": "inclusionAI/LLaDA-Image-Turbo", + "desc": "LLaDA-Image-Turbo is the distilled fast-generation and editing variant of LLaDA-Image.", + "extras": "steps: 4, cfg_scale: 1.0", + "size": 37.15, + "date": "2026 September" + }, + "SDXL Flash Mini": { + "path": "SDXL-Flash_Mini.safetensors@https://huggingface.co/sd-community/sdxl-flash-mini/resolve/main/SDXL-Flash_Mini.safetensors?download=true", + "preview": "SDXL-Flash_Mini.jpg", + "desc": "Introducing the new fast model SDXL Flash (Mini), we learned that all fast XL models work fast, but the quality decreases, and we also made a fast model, but it is not as fast as LCM, Turbo, Lightning and Hyper, but the quality is higher.", + "extras": "sampler: DEIS, steps: 40, cfg_scale: 6.0", + "experimental": true } } diff --git a/data/reference-nunchaku.json b/data/reference-nunchaku.json index be4ae4a76..17d49437e 100644 --- a/data/reference-nunchaku.json +++ b/data/reference-nunchaku.json @@ -1,4 +1,27 @@ { + "SDXL Base Nunchaku SVDQuant": { + "path": "stabilityai/stable-diffusion-xl-base-1.0", + "subfolder": "nunchaku", + "preview": "stabilityai--stable-diffusion-xl-base-1.0.jpg", + "desc": "Nunchaku SVDQuant quantization of SDXL Base 1.0 UNet with INT4 and SVD rank 32", + "nunchaku": [ + "Model" + ], + "size": 32.0, + "date": "2023 July" + }, + "SDXL Turbo Nunchaku SVDQuant": { + "path": "stabilityai/sdxl-turbo", + "subfolder": "nunchaku", + "preview": "stabilityai--sdxl-turbo.jpg", + "desc": "Nunchaku SVDQuant quantization of SDXL Turbo UNet with INT4 and SVD rank 32", + "nunchaku": [ + "Model" + ], + "extras": "sampler: Default, cfg_scale: 1.0, steps: 4", + "size": 19.38, + "date": "2023 November" + }, "FLUX.1-Dev Nunchaku SVDQuant": { "path": "black-forest-labs/FLUX.1-dev", "subfolder": "nunchaku", @@ -24,30 +47,6 @@ "size": 31.41, "date": "2024 July" }, - "FLUX.1-Kontext Nunchaku SVDQuant": { - "path": "black-forest-labs/FLUX.1-Kontext-dev", - "subfolder": "nunchaku", - "preview": "black-forest-labs--FLUX.1-Kontext-dev.jpg", - "desc": "Nunchaku SVDQuant quantization of FLUX.1-Kontext-dev transformer with INT4 and SVD rank 32", - "nunchaku": [ - "Model", - "TE" - ], - "size": 31.42, - "date": "2025 May" - }, - "FLUX.1-Krea Nunchaku SVDQuant": { - "path": "black-forest-labs/FLUX.1-Krea-dev", - "subfolder": "nunchaku", - "preview": "black-forest-labs--FLUX.1-Krea-dev.jpg", - "desc": "Nunchaku SVDQuant quantization of FLUX.1-Krea-dev transformer with INT4 and SVD rank 32", - "nunchaku": [ - "Model", - "TE" - ], - "size": 31.42, - "date": "2025 July" - }, "FLUX.1-Fill Nunchaku SVDQuant": { "path": "black-forest-labs/FLUX.1-Fill-dev", "subfolder": "nunchaku", @@ -74,6 +73,17 @@ "size": 40.68, "date": "2024 November" }, + "Sana 1.6B 1k Nunchaku SVDQuant": { + "path": "Efficient-Large-Model/Sana_1600M_1024px_BF16_diffusers", + "subfolder": "nunchaku", + "preview": "Efficient-Large-Model--Sana_1600M_1024px_diffusers.jpg", + "desc": "Nunchaku SVDQuant quantization of Sana 1.6B 1024px transformer with INT4 and SVD rank 32", + "nunchaku": [ + "Model" + ], + "size": 22.22, + "date": "2024 December" + }, "Shuttle Jaguar Nunchaku SVDQuant": { "path": "shuttleai/shuttle-jaguar", "subfolder": "nunchaku", @@ -86,6 +96,30 @@ "size": 31.41, "date": "2025 January" }, + "FLUX.1-Kontext Nunchaku SVDQuant": { + "path": "black-forest-labs/FLUX.1-Kontext-dev", + "subfolder": "nunchaku", + "preview": "black-forest-labs--FLUX.1-Kontext-dev.jpg", + "desc": "Nunchaku SVDQuant quantization of FLUX.1-Kontext-dev transformer with INT4 and SVD rank 32", + "nunchaku": [ + "Model", + "TE" + ], + "size": 31.42, + "date": "2025 May" + }, + "FLUX.1-Krea Nunchaku SVDQuant": { + "path": "black-forest-labs/FLUX.1-Krea-dev", + "subfolder": "nunchaku", + "preview": "black-forest-labs--FLUX.1-Krea-dev.jpg", + "desc": "Nunchaku SVDQuant quantization of FLUX.1-Krea-dev transformer with INT4 and SVD rank 32", + "nunchaku": [ + "Model", + "TE" + ], + "size": 31.42, + "date": "2025 July" + }, "Qwen-Image Nunchaku SVDQuant": { "path": "Qwen/Qwen-Image", "subfolder": "nunchaku", @@ -167,17 +201,6 @@ "size": 53.74, "date": "2025 September" }, - "Sana 1.6B 1k Nunchaku SVDQuant": { - "path": "Efficient-Large-Model/Sana_1600M_1024px_BF16_diffusers", - "subfolder": "nunchaku", - "preview": "Efficient-Large-Model--Sana_1600M_1024px_diffusers.jpg", - "desc": "Nunchaku SVDQuant quantization of Sana 1.6B 1024px transformer with INT4 and SVD rank 32", - "nunchaku": [ - "Model" - ], - "size": 22.22, - "date": "2024 December" - }, "Z-Image-Turbo Nunchaku SVDQuant": { "path": "Tongyi-MAI/Z-Image-Turbo", "subfolder": "nunchaku", @@ -190,29 +213,6 @@ "size": 30.58, "date": "2025 November" }, - "SDXL Base Nunchaku SVDQuant": { - "path": "stabilityai/stable-diffusion-xl-base-1.0", - "subfolder": "nunchaku", - "preview": "stabilityai--stable-diffusion-xl-base-1.0.jpg", - "desc": "Nunchaku SVDQuant quantization of SDXL Base 1.0 UNet with INT4 and SVD rank 32", - "nunchaku": [ - "Model" - ], - "size": 32.0, - "date": "2023 July" - }, - "SDXL Turbo Nunchaku SVDQuant": { - "path": "stabilityai/sdxl-turbo", - "subfolder": "nunchaku", - "preview": "stabilityai--sdxl-turbo.jpg", - "desc": "Nunchaku SVDQuant quantization of SDXL Turbo UNet with INT4 and SVD rank 32", - "nunchaku": [ - "Model" - ], - "extras": "sampler: Default, cfg_scale: 1.0, steps: 4", - "size": 19.38, - "date": "2023 November" - }, "Z-Image-Turbo Nunchaku-Lite": { "path": "lite-infer/z-image-turbo-nunchaku-lite-int4_r32-bnb4-text-encoder", "preview": "Tongyi-MAI--Z-Image-Turbo.jpg", diff --git a/data/reference-quantized.json b/data/reference-quantized.json index d9ccfbc70..ee0cbbd09 100644 --- a/data/reference-quantized.json +++ b/data/reference-quantized.json @@ -27,37 +27,6 @@ "size": 12.6, "date": "2025 October" }, - "FLUX.2 Dev sdnq-svd-uint4": { - "path": "Disty0/FLUX.2-dev-SDNQ-uint4-svd-r32", - "preview": "Disty0--FLUX.2-dev-SDNQ-uint4-svd-r32.jpg", - "desc": "Quantization of black-forest-labs/FLUX.2-dev using SDNQ: sdnq-svd 4-bit uint with svd rank 32", - "size": 31.89, - "date": "2025 November" - }, - "Black Forest Labs FLUX.2 Klein 4B sdnq-uint4-dynamic": { - "path": "Disty0/FLUX.2-klein-4B-SDNQ-4bit-dynamic", - "preview": "Disty0--FLUX.2-klein-4B-SDNQ-4bit-dynamic.jpg", - "desc": "Dynamic 4-bit quantization of black-forest-labs/FLUX.2-klein-4B using SDNQ.", - "extras": "sampler: Default, cfg_scale: 1.0, steps: 4", - "size": 5.09, - "date": "2026 January" - }, - "Black Forest Labs FLUX.2 Klein 9B sdnq-uint4-dynamic-svd": { - "path": "Disty0/FLUX.2-klein-9B-SDNQ-4bit-dynamic-svd-r32", - "preview": "Disty0--FLUX.2-klein-9B-SDNQ-4bit-dynamic-svd-r32.jpg", - "desc": "Dynamic 4-bit quantization of black-forest-labs/FLUX.2-klein-9B using SDNQ with SVD rank 32.", - "extras": "sampler: Default, cfg_scale: 1.0, steps: 4", - "size": 11.73, - "date": "2026 January" - }, - "Black Forest Labs FLUX.2 Klein 9B KV sdnq-uint4-dynamic-svd": { - "path": "vladmandic/Flux.2-Klein-9B-KV-sdnq-hadamard-uint4", - "preview": "black-forest-labs--FLUX.2-klein-9b-kv.jpg", - "desc": "Dynamic 4-bit quantization of black-forest-labs/FLUX.2-klein-9B-KV using SDNQ with Hadamard.", - "extras": "sampler: Default, cfg_scale: 1.0, steps: 4", - "size": 12.26, - "date": "2026 July" - }, "Chroma1-HD sdnq-svd-uint4": { "path": "Disty0/Chroma1-HD-SDNQ-uint4-svd-r32", "preview": "Disty0--Chroma1-HD-SDNQ-uint4-svd-r32.jpg", @@ -79,48 +48,6 @@ "date": "2025 October", "size": 23.53 }, - "MiniMaxAI MiniMax-H3 sdnq-uint4": { - "path": "OzzyGT/MiniMax_H3_sdnq_dynamic_4bit", - "preview": "MiniMaxAI--MiniMax-H3.jpg", - "desc": "Quantization of MiniMaxAI/MiniMax-H3 using SDNQ: dynamic 4-bit uint. Video with synchronized audio; in image tabs the model runs in experimental still mode.", - "extras": "sampler: Default", - "size": 64.80, - "date": "2026 August" - }, - "MiniMaxAI MiniMax-H3 sdnq-uint4 Ref2VA": { - "path": "OzzyGT/MiniMax_H3_sdnq_dynamic_4bit", - "preview": "MiniMaxAI--MiniMax-H3.jpg", - "subfolder": "ref2va", - "desc": "Quantization of MiniMaxAI/MiniMax-H3 using SDNQ: dynamic 4-bit uint. Video with synchronized audio; in image tabs the model runs in experimental still mode.", - "extras": "sampler: Default", - "size": 64.80, - "date": "2026 August" - }, - "MiniMaxAI MiniMax-H3 Pruned sdnq-uint4": { - "path": "OzzyGT/MiniMax_H3_sdnq_4bit_pruned", - "preview": "OzzyGT--MiniMax_H3_sdnq_4bit_pruned.jpg", - "desc": "Quantization of MiniMaxAI/MiniMax-H3 using SDNQ: dynamic 4-bit uint. Video with synchronized audio; in image tabs the model runs in experimental still mode.", - "extras": "sampler: Default", - "size": 23.70, - "date": "2026 August" - }, - "MiniMaxAI MiniMax-H3 Pruned sdnq-uint4 Ref2VA": { - "path": "OzzyGT/MiniMax_H3_sdnq_4bit_pruned", - "preview": "OzzyGT--MiniMax_H3_sdnq_4bit_pruned.jpg", - "subfolder": "ref2va", - "desc": "Quantization of MiniMaxAI/MiniMax-H3 using SDNQ: dynamic 4-bit uint. Video with synchronized audio; in image tabs the model runs in experimental still mode.", - "extras": "sampler: Default", - "size": 23.70, - "date": "2026 August" - }, - "Z-Image-Turbo sdnq-svd-uint4": { - "path": "Disty0/Z-Image-Turbo-SDNQ-uint4-svd-r32", - "preview": "Disty0--Z-Image-Turbo-SDNQ-uint4-svd-r32.jpg", - "desc": "Quantization of Tongyi-MAI/Z-Image-Turbo using SDNQ: sdnq-svd 4-bit uint with svd rank 32", - "extras": "sampler: Default, cfg_scale: 1.0, steps: 9", - "size": 6.05, - "date": "2025 November" - }, "Qwen-Image sdnq-svd-uint4": { "path": "Disty0/Qwen-Image-SDNQ-uint4-svd-r32", "preview": "Qwen--Qwen-Image.jpg", @@ -128,13 +55,6 @@ "date": "2025 October", "size": 16.09 }, - "Qwen-Image-2512 sdnq-svd-uint4": { - "path": "Disty0/Qwen-Image-2512-SDNQ-uint4-svd-r32", - "preview": "Disty0--Qwen-Image-2512-SDNQ-uint4-svd-r32.jpg", - "desc": "Quantization of Qwen/Qwen-Image-2512 using SDNQ: sdnq-svd 4-bit uint with svd rank 32", - "size": 16.09, - "date": "2026 January" - }, "Qwen-Image-Edit sdnq-svd-uint4": { "path": "Disty0/Qwen-Image-Edit-SDNQ-uint4-svd-r32", "preview": "Qwen--Qwen-Image-Edit.jpg", @@ -149,20 +69,6 @@ "date": "2025 October", "size": 16.09 }, - "Qwen-Image-Edit-2511 sdnq-svd-uint4": { - "path": "Disty0/Qwen-Image-Edit-2511-SDNQ-uint4-svd-r32", - "preview": "Disty0--Qwen-Image-Edit-2511-SDNQ-uint4-svd-r32.jpg", - "desc": "Quantization of Qwen/Qwen-Image-Edit-2511 using SDNQ: sdnq-svd 4-bit uint with svd rank 32", - "date": "2025 December", - "size": 16.09 - }, - "Qwen-Image-Layered sdnq-svd-uint4": { - "path": "Disty0/Qwen-Image-Layered-SDNQ-uint4-svd-r32", - "preview": "Disty0--Qwen-Image-Layered-SDNQ-uint4-svd-r32.jpg", - "desc": "Quantization of Qwen/Qwen-Image-Layered using SDNQ: sdnq-svd 4-bit uint with svd rank 32", - "date": "2025 December", - "size": 16.09 - }, "nVidia ChronoEdit sdnq-svd-uint4": { "path": "Disty0/ChronoEdit-14B-SDNQ-uint4-svd-r32", "preview": "Disty0--ChronoEdit-14B-SDNQ-uint4-svd-r32.jpg", @@ -198,6 +104,58 @@ "size": 3.37, "date": "2025 October" }, + "FLUX.2 Dev sdnq-svd-uint4": { + "path": "Disty0/FLUX.2-dev-SDNQ-uint4-svd-r32", + "preview": "Disty0--FLUX.2-dev-SDNQ-uint4-svd-r32.jpg", + "desc": "Quantization of black-forest-labs/FLUX.2-dev using SDNQ: sdnq-svd 4-bit uint with svd rank 32", + "size": 31.89, + "date": "2025 November" + }, + "Z-Image-Turbo sdnq-svd-uint4": { + "path": "Disty0/Z-Image-Turbo-SDNQ-uint4-svd-r32", + "preview": "Disty0--Z-Image-Turbo-SDNQ-uint4-svd-r32.jpg", + "desc": "Quantization of Tongyi-MAI/Z-Image-Turbo using SDNQ: sdnq-svd 4-bit uint with svd rank 32", + "extras": "sampler: Default, cfg_scale: 1.0, steps: 9", + "size": 6.05, + "date": "2025 November" + }, + "Qwen-Image-Edit-2511 sdnq-svd-uint4": { + "path": "Disty0/Qwen-Image-Edit-2511-SDNQ-uint4-svd-r32", + "preview": "Disty0--Qwen-Image-Edit-2511-SDNQ-uint4-svd-r32.jpg", + "desc": "Quantization of Qwen/Qwen-Image-Edit-2511 using SDNQ: sdnq-svd 4-bit uint with svd rank 32", + "date": "2025 December", + "size": 16.09 + }, + "Qwen-Image-Layered sdnq-svd-uint4": { + "path": "Disty0/Qwen-Image-Layered-SDNQ-uint4-svd-r32", + "preview": "Disty0--Qwen-Image-Layered-SDNQ-uint4-svd-r32.jpg", + "desc": "Quantization of Qwen/Qwen-Image-Layered using SDNQ: sdnq-svd 4-bit uint with svd rank 32", + "date": "2025 December", + "size": 16.09 + }, + "Black Forest Labs FLUX.2 Klein 4B sdnq-uint4-dynamic": { + "path": "Disty0/FLUX.2-klein-4B-SDNQ-4bit-dynamic", + "preview": "Disty0--FLUX.2-klein-4B-SDNQ-4bit-dynamic.jpg", + "desc": "Dynamic 4-bit quantization of black-forest-labs/FLUX.2-klein-4B using SDNQ.", + "extras": "sampler: Default, cfg_scale: 1.0, steps: 4", + "size": 5.09, + "date": "2026 January" + }, + "Black Forest Labs FLUX.2 Klein 9B sdnq-uint4-dynamic-svd": { + "path": "Disty0/FLUX.2-klein-9B-SDNQ-4bit-dynamic-svd-r32", + "preview": "Disty0--FLUX.2-klein-9B-SDNQ-4bit-dynamic-svd-r32.jpg", + "desc": "Dynamic 4-bit quantization of black-forest-labs/FLUX.2-klein-9B using SDNQ with SVD rank 32.", + "extras": "sampler: Default, cfg_scale: 1.0, steps: 4", + "size": 11.73, + "date": "2026 January" + }, + "Qwen-Image-2512 sdnq-svd-uint4": { + "path": "Disty0/Qwen-Image-2512-SDNQ-uint4-svd-r32", + "preview": "Disty0--Qwen-Image-2512-SDNQ-uint4-svd-r32.jpg", + "desc": "Quantization of Qwen/Qwen-Image-2512 using SDNQ: sdnq-svd 4-bit uint with svd rank 32", + "size": 16.09, + "date": "2026 January" + }, "ZAI GLM-Image sdnq-dynamic-uint4": { "path": "Disty0/GLM-Image-SDNQ-4bit-dynamic", "preview": "zai-org--GLM-Image.jpg", @@ -260,6 +218,22 @@ "size": 17.3, "date": "2026 June" }, + "Krea 2 Base sdnq-hadamard-uint4": { + "path": "vladmandic/Krea-2-Base-sdnq-hadamard-uint4", + "preview": "CalamitousFelicitousness--Krea-2-Base-Diffusers.jpg", + "desc": "Krea 2 (K2) Base is the undistilled foundation model of the Krea 2 family, trained from scratch by Krea. A 12.9B-parameter single-stream flow-matching DiT that uses a Qwen3-VL-4B vision-language model as its text encoder and the Qwen-Image VAE. The base checkpoint is intended for fine-tuning and LoRA training; LoRAs trained on it apply to Krea 2 Turbo.", + "extras": "sampler: Default, cfg_scale: 4.5, steps: 52", + "size": 10.3, + "date": "2026 June" + }, + "Black Forest Labs FLUX.2 Klein 9B KV sdnq-uint4-dynamic-svd": { + "path": "vladmandic/Flux.2-Klein-9B-KV-sdnq-hadamard-uint4", + "preview": "black-forest-labs--FLUX.2-klein-9b-kv.jpg", + "desc": "Dynamic 4-bit quantization of black-forest-labs/FLUX.2-klein-9B-KV using SDNQ with Hadamard.", + "extras": "sampler: Default, cfg_scale: 1.0, steps: 4", + "size": 12.26, + "date": "2026 July" + }, "Krea 2 Turbo sdnq-hadamard-uint4": { "path": "vladmandic/Krea-2-Turbo-sdnq-hadamard-uint4", "preview": "CalamitousFelicitousness--Krea-2-Turbo-Diffusers.jpg", @@ -268,12 +242,38 @@ "size": 10.54, "date": "2026 July" }, - "Krea 2 Base sdnq-hadamard-uint4": { - "path": "vladmandic/Krea-2-Base-sdnq-hadamard-uint4", - "preview": "CalamitousFelicitousness--Krea-2-Base-Diffusers.jpg", - "desc": "Krea 2 (K2) Base is the undistilled foundation model of the Krea 2 family, trained from scratch by Krea. A 12.9B-parameter single-stream flow-matching DiT that uses a Qwen3-VL-4B vision-language model as its text encoder and the Qwen-Image VAE. The base checkpoint is intended for fine-tuning and LoRA training; LoRAs trained on it apply to Krea 2 Turbo.", - "extras": "sampler: Default, cfg_scale: 4.5, steps: 52", - "size": 10.3, - "date": "2026 June" + "MiniMaxAI MiniMax-H3 sdnq-uint4": { + "path": "OzzyGT/MiniMax_H3_sdnq_dynamic_4bit", + "preview": "MiniMaxAI--MiniMax-H3.jpg", + "desc": "Quantization of MiniMaxAI/MiniMax-H3 using SDNQ: dynamic 4-bit uint. Video with synchronized audio; in image tabs the model runs in experimental still mode.", + "extras": "sampler: Default", + "size": 64.80, + "date": "2026 August" + }, + "MiniMaxAI MiniMax-H3 sdnq-uint4 Ref2VA": { + "path": "OzzyGT/MiniMax_H3_sdnq_dynamic_4bit", + "preview": "MiniMaxAI--MiniMax-H3.jpg", + "subfolder": "ref2va", + "desc": "Quantization of MiniMaxAI/MiniMax-H3 using SDNQ: dynamic 4-bit uint. Video with synchronized audio; in image tabs the model runs in experimental still mode.", + "extras": "sampler: Default", + "size": 64.80, + "date": "2026 August" + }, + "MiniMaxAI MiniMax-H3 Pruned sdnq-uint4": { + "path": "OzzyGT/MiniMax_H3_sdnq_4bit_pruned", + "preview": "OzzyGT--MiniMax_H3_sdnq_4bit_pruned.jpg", + "desc": "Quantization of MiniMaxAI/MiniMax-H3 using SDNQ: dynamic 4-bit uint. Video with synchronized audio; in image tabs the model runs in experimental still mode.", + "extras": "sampler: Default", + "size": 23.70, + "date": "2026 August" + }, + "MiniMaxAI MiniMax-H3 Pruned sdnq-uint4 Ref2VA": { + "path": "OzzyGT/MiniMax_H3_sdnq_4bit_pruned", + "preview": "OzzyGT--MiniMax_H3_sdnq_4bit_pruned.jpg", + "subfolder": "ref2va", + "desc": "Quantization of MiniMaxAI/MiniMax-H3 using SDNQ: dynamic 4-bit uint. Video with synchronized audio; in image tabs the model runs in experimental still mode.", + "extras": "sampler: Default", + "size": 23.70, + "date": "2026 August" } }