diff --git a/README.md b/README.md
index 2518a82..2f482dc 100644
--- a/README.md
+++ b/README.md
@@ -29,6 +29,7 @@
 项目只读取一个配置文件：`config.json`。
 
 - `services.openai.host` / `services.openai.port`：OpenAI 协议服务监听地址与端口（默认 `0.0.0.0:8001`）
+- `public_model_name`：对外固定模型名，切换底层模型时可保持调用方参数不变
 - `api_key`：OpenAI 接口访问密钥
 - `tensor_parallel_size`：张量并行数，双卡建议 `2`
 - `dtype`：推理精度，默认 `bfloat16`
@@ -82,14 +83,14 @@ curl http://localhost:<services.openai.port>/v1/models
 curl -X POST "http://localhost:8001/v1/chat/completions" \
   -H "Content-Type: application/json" \
   -H "Authorization: Bearer <config.json中的api_key>" \
-  -d "{\"model\":\"Qwen3.5-35B-A3B-GPTQ-Int4\",\"messages\":[{\"role\":\"user\",\"content\":\"你好，介绍一下你自己\"}],\"temperature\":0.7}"
+  -d "{\"model\":\"Qwen_local_model\",\"messages\":[{\"role\":\"user\",\"content\":\"你好，介绍一下你自己\"}],\"temperature\":0.7}"
 ```
 
 ## OpenClaw 调用说明
 
 - Base URL 使用 `http://<服务器IP>:8001/v1`
 - API Key 使用 `config.json` 中 `api_key`
-- 模型名使用 `config.json` 中 `models.profiles.<模型名>.served_model_name`
+- 模型名固定使用 `config.json` 中 `public_model_name`（默认 `Qwen_local_model`）
 - 若使用工具调用，`config.json` 中应配置 `tool_call_parser` 与 `enable_auto_tool_choice`
 - 服务强制离线模式，不会回退到 Hugging Face 远程下载
 - 所有路径按 Ubuntu 规范填写，本地模型建议使用 `/opt/model/<模型目录>`
@@ -105,5 +106,6 @@ curl -X POST "http://localhost:8001/v1/chat/completions" \
 
 - 报错 `model type ... Transformers does not recognize this architecture` 时，说明当前模型与镜像内依赖不兼容，建议更换模型或升级镜像版本。
 - 报错 `model config (gptq) does not match quantization argument (gptq_marlin)` 时，将该模型配置改为 `dtype=float16` 且 `quantization=gptq`。
+- 报错 `RPC call to sample_tokens timed out` 或出现 `GPU core dump` 时，先下调模型配置为更稳参数：`ctx=32768`、`max_num_seqs=4`、`max_tokens=2048`、`gpu_util=0.90`，并开启 `enforce_eager=true`。
 - 若模型目录存在但仍加载失败，检查挂载路径是否为 `/opt/model:/opt/model:ro`，并确认容器内可见模型文件。
 - 如果看到 `No services to build`，说明未触发重建；需要先执行 `docker compose build --no-cache` 再 `up`。
diff --git a/app/config.py b/app/config.py
index 821d65e..0d98a35 100644
--- a/app/config.py
+++ b/app/config.py
@@ -17,6 +17,7 @@ class Settings(BaseModel):
     port: int = 8000
     openai_host: str = "0.0.0.0"
     openai_port: int = 8001
+    public_model_name: str = "Qwen_local_model"
     model_root: str = "/opt/model"
     offline_mode: bool = True
     max_model_len: int = 8192
@@ -44,6 +45,7 @@ def get_settings() -> Settings:
         port=runtime["port"],
         openai_host=runtime["openai_host"],
         openai_port=runtime["openai_port"],
+        public_model_name=runtime["public_model_name"],
         model_root=runtime["model_root"],
         offline_mode=runtime["offline_mode"],
         api_key=runtime["api_key"],
diff --git a/app/model_catalog.py b/app/model_catalog.py
index 75fc3e8..3a3c3ba 100644
--- a/app/model_catalog.py
+++ b/app/model_catalog.py
@@ -68,6 +68,7 @@ def resolve_runtime_settings(content: dict[str, Any]) -> dict[str, Any]:
         "port": _to_int(api_service.get("port"), 8000),
         "openai_host": str(openai_service.get("host", "0.0.0.0")),
         "openai_port": _to_int(openai_service.get("port"), 8001),
+        "public_model_name": _to_str(content.get("public_model_name"), "Qwen_local_model"),
         "api_key": str(content.get("api_key", "")).strip() or None,
         "tensor_parallel_size": _to_int(content.get("tensor_parallel_size"), 2),
         "dtype": str(content.get("dtype", "bfloat16")),
diff --git a/app/start_openai.py b/app/start_openai.py
index 6df4540..b32479f 100644
--- a/app/start_openai.py
+++ b/app/start_openai.py
@@ -18,6 +18,7 @@ def build_command() -> list[str]:
         os.environ[key] = value
     host = str(runtime["openai_host"])
     port = str(runtime["openai_port"])
+    public_model_name = str(runtime["public_model_name"]).strip()
     api_key = runtime["api_key"] or ""
     dtype = str(updates["dtype"] or runtime["dtype"])
     quantization = str(updates["quantization"] or "").strip()
@@ -33,7 +34,7 @@ def build_command() -> list[str]:
         "--model",
         str(updates["model_name"]),
         "--served-model-name",
-        str(updates["served_model_name"]),
+        public_model_name or str(updates["served_model_name"]),
         "--tensor-parallel-size",
         str(updates["tensor_parallel_size"]),
         "--max-model-len",
diff --git a/config.json b/config.json
index d682453..fac72a3 100644
--- a/config.json
+++ b/config.json
@@ -9,6 +9,7 @@
       "port": 8001
     }
   },
+  "public_model_name": "Qwen_local_model",
   "api_key": "sk-szcjw",
   "tensor_parallel_size": 2,
   "dtype": "bfloat16",
@@ -70,9 +71,10 @@
         "ctx": "65536",
         "trust_remote": true,
         "valid_tp": [1, 2],
-        "max_num_seqs": "8",
-        "max_tokens": "8192",
-        "gpu_util": "0.92",
+        "max_num_seqs": "4",
+        "max_tokens": "4096",
+        "gpu_util": "0.94",
+        "enforce_eager": true,
         "tool_call_parser": "qwen3_xml",
         "enable_auto_tool_choice": true,
         "served_model_name": "Qwen3.5-35B-A3B-GPTQ-Int4",