feat(chat): expose model options config

README.md

@@ -269,18 +269,39 @@ docker compose down
 deepseek/deepseek-v4-flash
 ```
 
+默认聊天模型配置组为：
+
+```json
+[
+  {
+    "id": "deepseek/deepseek-v4-flash",
+    "label": "快速",
+    "description": "快速回答和任务执行",
+    "icon": "bolt"
+  },
+  {
+    "id": "deepseek/deepseek-v4-pro",
+    "label": "专家",
+    "description": "探索、解决复杂任务",
+    "icon": "sparkle"
+  }
+]
+```
+
 涉及位置：
 
 ```text
-opencode.json
-.opencode/agents/tjwater-assistant.md
-src/config.ts 的 OPENCODE_MODEL 默认值
+src/chat/modelConfig.ts 的默认模型配置组
+src/config.ts 的 OPENCODE_MODEL 与 OPENCODE_MODEL_OPTIONS 默认值
+opencode.json 的 opencode 运行时默认模型
 ```
 
-如果需要临时覆盖模型，可以在启动时设置：
+如果需要临时覆盖默认模型和模型配置组，可以在启动时设置：
 
 ```bash
-OPENCODE_MODEL=deepseek/deepseek-v4-pro bun run start
+OPENCODE_MODEL=deepseek/deepseek-v4-pro \
+OPENCODE_MODEL_OPTIONS='[{"id":"deepseek/deepseek-v4-flash","label":"快速","description":"快速回答和任务执行","icon":"bolt"},{"id":"deepseek/deepseek-v4-pro","label":"专家","description":"探索、解决复杂任务","icon":"sparkle"}]' \
+bun run start
 ```
 
 DeepSeek API key 不写入代码，部署时通过环境变量设置：