Add Qwen3.6 validation and benchmark coverage

tools/bench/client.py

@@ -110,12 +110,16 @@ async def chat_stream(
                 choice = choices[0]
                 delta = choice.get("delta") or {}
                 content = delta.get("content")
-                if content:
+                reasoning_content = delta.get("reasoning_content")
+                token_text = content if content is not None else reasoning_content
+                # Exclude the role announcement, but count every generated-token
+                # chunk, including an empty UTF-8 fragment, in TTFT/TPOT.
+                if token_text is not None and "role" not in delta:
                     now = time.perf_counter()
                     if res.ttft_s < 0:
                         res.ttft_s = now - t_start
                     res.chunk_times.append(now)
-                    res.text += content
+                    res.text += token_text
                 if choice.get("finish_reason"):
                     res.finish_reason = choice["finish_reason"]
     except Exception as e:  # noqa: BLE001 — surface any failure to the report