diff --git a/tools/server/server-common.h b/tools/server/server-common.h
index b782130a25..52b4df4c61 100644
--- a/tools/server/server-common.h
+++ b/tools/server/server-common.h
@@ -338,8 +338,6 @@ json format_response_rerank(
 // stats and metrics
 //
 
-struct server_slot;
-
 // shared between server_slot and server_task_result_*
 struct server_slot_stats {
     uint64_t n_prompt_cached    = 0;
@@ -507,8 +505,6 @@ struct server_metrics {
         n_prompt_queued = 0;
     }
 
-    // this is implemented in server-context.cpp
-    void on_prediction(const server_slot & slot);
 };
 
 //
diff --git a/tools/server/server-context.cpp b/tools/server/server-context.cpp
index e3958637d3..6b0eade736 100644
--- a/tools/server/server-context.cpp
+++ b/tools/server/server-context.cpp
@@ -780,30 +780,6 @@ struct server_slot {
 };
 
 
-//
-// server_metrics
-//
-
-void server_metrics::on_prediction(const server_slot & slot) {
-    const uint64_t t_us    = slot.stats.t_gen_us();
-    const uint64_t n       = slot.stats.n_gen;
-    const uint64_t n_steps = slot.stats.n_gen_steps();
-
-    predict       .add(n, n_steps, t_us);
-    predict_bucket.add(n, n_steps, t_us);
-
-    n_draft_tokens      += slot.stats.n_draft_tokens;
-    n_draft_accepted    += slot.stats.n_draft_accepted;
-    n_draft_verif_steps += slot.stats.n_draft_verif_steps;
-
-    if (n_accepted_per_pos.size() < slot.n_accepted_per_pos.size()) {
-        n_accepted_per_pos.resize(slot.n_accepted_per_pos.size(), 0);
-    }
-    for (size_t i = 0; i < slot.n_accepted_per_pos.size(); i++) {
-        n_accepted_per_pos[i] += slot.n_accepted_per_pos[i];
-    }
-}
-
 
 //
 // server_context_impl (private implementation)
@@ -3762,7 +3738,7 @@ private:
                 // release slot because of stop condition
                 slot.print_timings();
                 send_final_response(slot);
-                metrics.on_prediction(slot);
+                metrics_on_prediction(slot);
                 slot.release();
 
                 return;
@@ -3883,7 +3859,7 @@ private:
                 if (!process_token(result, slot)) {
                     slot.print_timings();
                     send_final_response(slot);
-                    metrics.on_prediction(slot);
+                    metrics_on_prediction(slot);
                     slot.release();
 
                     return;
@@ -3945,6 +3921,29 @@ private:
             metrics.flush_prompt();
         }
     }
+
+    void metrics_on_prediction(const server_slot & slot) {
+        const uint64_t t_us    = slot.stats.t_gen_us();
+        const uint64_t n       = slot.stats.n_gen;
+        const uint64_t n_steps = slot.stats.n_gen_steps();
+
+        metrics.predict       .add(n, n_steps, t_us);
+        metrics.predict_bucket.add(n, n_steps, t_us);
+
+        metrics.n_draft_tokens      += slot.stats.n_draft_tokens;
+        metrics.n_draft_accepted    += slot.stats.n_draft_accepted;
+        metrics.n_draft_verif_steps += slot.stats.n_draft_verif_steps;
+
+        auto & dst = metrics.n_accepted_per_pos;
+        const auto & src = slot.n_accepted_per_pos;
+
+        if (dst.size() < src.size()) {
+            dst.resize(src.size(), 0);
+        }
+        for (size_t i = 0; i < src.size(); i++) {
+            dst[i] += src[i];
+        }
+    }
 };
 
 //