addd parameters

clusters/k3s-dgx/nim-service/qwen.yaml

@@ -55,19 +55,34 @@ spec:
 
   command:
     - python3
-
   args:
     - -m
     - vllm.entrypoints.openai.api_server
     - --model
     - /model-store
+    - --host
+    - 0.0.0.0
     - --served-model-name
     - qwen36
     - --gpu-memory-utilization
     - "0.85"
     - --max-model-len
-    - "262144"
+    - "256000"
+    - --language-model-only
+    - --reasoning-parser
+    - qwen3
+    - --enable-auto-tool-choice
+    - --tool-call-parser
+    - qwen3_coder
+    - --enable-chunked-prefill
+    - --max-num-batched-tokens
+    - "32768"
+    - --max-num-seqs
+    - "10"
     - --enable-prefix-caching
+    - --speculative-config
+    - '{"method":"mtp","num_speculative_tokens":2}'
+
   authSecret: hf-api-secret
   storage:
     nimCache: