diff --git a/README.md b/README.md
index 724fd594d6..1a80a18b7b 100644
--- a/README.md
+++ b/README.md
@@ -293,7 +293,7 @@ Then open **http://127.0.0.1:8088/** for the Console. Config, memory, and skills
 >   -v copaw-secrets:/app/working.secret \
 >   agentscope/copaw:latest
 > ```
-> Then in CoPaw **Settings → Models → Ollama**, change the Base URL to `http://host.docker.internal:11434` or your corresponding port.
+> Then in CoPaw **Settings → Models**, change the Base URL to `http://host.docker.internal:<port>` — for example, `http://host.docker.internal:11434` for Ollama, or `http://host.docker.internal:1234/v1` for LM Studio.
 >
 > **Option B** — Host networking (Linux only):
 > ```bash
diff --git a/README_zh.md b/README_zh.md
index fed00f3085..b189a5a02e 100644
--- a/README_zh.md
+++ b/README_zh.md
@@ -295,7 +295,7 @@ docker run -p 127.0.0.1:8088:8088 \
 >   -v copaw-secrets:/app/working.secret \
 >   agentscope/copaw:latest
 > ```
-> 然后在 CoPaw **设置 → 模型 → Ollama** 中，将 Base URL 改为 `http://host.docker.internal:11434` 或对应端口。
+> 然后在 CoPaw **设置 → 模型** 中，将 Base URL 改为 `http://host.docker.internal:<端口>` — 例如 Ollama 填 `http://host.docker.internal:11434`，LM Studio 填 `http://host.docker.internal:1234/v1`。
 >
 > **方式 B** — 使用宿主机网络（仅限 Linux）：
 > ```bash
diff --git a/console/src/locales/en.json b/console/src/locales/en.json
index 5306816372..052516f182 100644
--- a/console/src/locales/en.json
+++ b/console/src/locales/en.json
@@ -353,6 +353,7 @@
     "azureEndpointHint": "Azure OpenAI endpoint, e.g. https://<resource>.openai.azure.com/openai/v1",
     "anthropicEndpointHint": "Anthropic endpoint, e.g. https://api.anthropic.com",
     "ollamaEndpointHint": "Ollama endpoint, e.g. http://localhost:11434",
+    "lmstudioEndpointHint": "LM Studio endpoint, e.g. http://localhost:1234/v1",
     "apiEndpointHint": "API endpoint, e.g. https://api.example.com",
     "pleaseEnterBaseURL": "Please enter the API base URL",
     "pleaseEnterValidURL": "Please enter a valid URL",
diff --git a/console/src/locales/ja.json b/console/src/locales/ja.json
index 69176f2cf1..4a8be4f2d6 100644
--- a/console/src/locales/ja.json
+++ b/console/src/locales/ja.json
@@ -350,6 +350,7 @@
     "azureEndpointHint": "Azure OpenAIエンドポイント、例: https://<resource>.openai.azure.com/openai/v1",
     "anthropicEndpointHint": "Anthropicエンドポイント、例: https://api.anthropic.com",
     "ollamaEndpointHint": "Ollamaエンドポイント、例: http://localhost:11434",
+    "lmstudioEndpointHint": "LM Studioエンドポイント、例: http://localhost:1234/v1",
     "apiEndpointHint": "APIエンドポイント、例: https://api.example.com",
     "pleaseEnterBaseURL": "APIベースURLを入力してください",
     "pleaseEnterValidURL": "有効なURLを入力してください",
diff --git a/console/src/locales/ru.json b/console/src/locales/ru.json
index b066b3927b..c89df2c4be 100644
--- a/console/src/locales/ru.json
+++ b/console/src/locales/ru.json
@@ -353,6 +353,7 @@
     "azureEndpointHint": "Azure OpenAI endpoint, например https://<resource>.openai.azure.com/openai/v1",
     "anthropicEndpointHint": "Anthropic endpoint, например https://api.anthropic.com",
     "ollamaEndpointHint": "Ollama endpoint, например http://localhost:11434",
+    "lmstudioEndpointHint": "LM Studio endpoint, например http://localhost:1234/v1",
     "apiEndpointHint": "API endpoint, например https://api.example.com",
     "pleaseEnterBaseURL": "Пожалуйста, введите базовый URL API",
     "pleaseEnterValidURL": "Пожалуйста, введите корректный URL",
diff --git a/console/src/locales/zh.json b/console/src/locales/zh.json
index 3a7d6c3de2..e17587c3f5 100644
--- a/console/src/locales/zh.json
+++ b/console/src/locales/zh.json
@@ -353,6 +353,7 @@
     "azureEndpointHint": "Azure OpenAI 端点，例如 https://<resource>.openai.azure.com/openai/v1",
     "anthropicEndpointHint": "Anthropic 端点，例如 https://api.anthropic.com",
     "ollamaEndpointHint": "Ollama 端点，例如 http://localhost:11434",
+    "lmstudioEndpointHint": "LM Studio 端点，例如 http://localhost:1234/v1",
     "apiEndpointHint": "API 端点，例如 https://api.example.com",
     "pleaseEnterBaseURL": "请输入 API 基础 URL",
     "pleaseEnterValidURL": "请输入有效的 URL",
diff --git a/console/src/pages/Settings/Models/components/cards/RemoteProviderCard.tsx b/console/src/pages/Settings/Models/components/cards/RemoteProviderCard.tsx
index a9ae620507..4157dbc7e4 100644
--- a/console/src/pages/Settings/Models/components/cards/RemoteProviderCard.tsx
+++ b/console/src/pages/Settings/Models/components/cards/RemoteProviderCard.tsx
@@ -66,7 +66,6 @@ export function RemoteProviderCard({
   } else if (provider.is_custom && provider.base_url) {
     isConfigured = true;
   } else if (provider.require_api_key === false) {
-    // If API key is not required, consider it configured
     isConfigured = true;
   } else if (provider.require_api_key && provider.api_key) {
     isConfigured = true;
diff --git a/console/src/pages/Settings/Models/components/modals/ProviderConfigModal.tsx b/console/src/pages/Settings/Models/components/modals/ProviderConfigModal.tsx
index d0f30f433e..7adac809a0 100644
--- a/console/src/pages/Settings/Models/components/modals/ProviderConfigModal.tsx
+++ b/console/src/pages/Settings/Models/components/modals/ProviderConfigModal.tsx
@@ -328,6 +328,9 @@ export function ProviderConfigModal({
     if (provider.id === "ollama") {
       return t("models.ollamaEndpointHint");
     }
+    if (provider.id === "lmstudio") {
+      return t("models.lmstudioEndpointHint");
+    }
     if (provider.is_custom) {
       return effectiveChatModel === "AnthropicChatModel"
         ? t("models.anthropicEndpointHint")
@@ -352,6 +355,9 @@ export function ProviderConfigModal({
     if (provider.id === "ollama") {
       return "http://localhost:11434";
     }
+    if (provider.id === "lmstudio") {
+      return "http://localhost:1234/v1";
+    }
     if (provider.is_custom && effectiveChatModel === "AnthropicChatModel") {
       return "https://api.anthropic.com";
     }
diff --git a/console/src/pages/Settings/Models/components/sections/ModelsSection.tsx b/console/src/pages/Settings/Models/components/sections/ModelsSection.tsx
index ce8e3ddd0f..875deb41c1 100644
--- a/console/src/pages/Settings/Models/components/sections/ModelsSection.tsx
+++ b/console/src/pages/Settings/Models/components/sections/ModelsSection.tsx
@@ -51,7 +51,7 @@ export function ModelsSection({
           (p.models?.length ?? 0) + (p.extra_models?.length ?? 0) > 0;
         if (!hasModels) return false;
         if (p.is_local) return true;
-        if (p.id === "ollama") return !!p.base_url;
+        if (p.require_api_key === false) return !!p.base_url;
         if (p.is_custom) return !!p.base_url;
         if (p.require_api_key ?? true) return !!p.api_key;
         return true;
diff --git a/src/copaw/providers/lm_studio_provider.py b/src/copaw/providers/lm_studio_provider.py
new file mode 100644
index 0000000000..2e80e31a0f
--- /dev/null
+++ b/src/copaw/providers/lm_studio_provider.py
@@ -0,0 +1,27 @@
+# -*- coding: utf-8 -*-
+"""An LM Studio provider implementation.
+
+LM Studio exposes an OpenAI-compatible local server (default
+http://localhost:1234/v1). This provider auto-discovers loaded models
+on each get_info() call, similar to OllamaProvider."""
+
+from __future__ import annotations
+
+import logging
+
+from copaw.providers.openai_provider import OpenAIProvider
+from copaw.providers.provider import ProviderInfo
+
+logger = logging.getLogger(__name__)
+
+
+class LMStudioProvider(OpenAIProvider):
+    """Provider for LM Studio's OpenAI-compatible local server."""
+
+    async def get_info(self, mock_secret: bool = True) -> ProviderInfo:
+        try:
+            models = await self.fetch_models(timeout=1)
+            self.models = models
+        except Exception as exc:
+            logger.debug("LM Studio model discovery failed: %s", exc)
+        return await super().get_info(mock_secret=mock_secret)
diff --git a/src/copaw/providers/provider_manager.py b/src/copaw/providers/provider_manager.py
index b707d8e208..7545031834 100644
--- a/src/copaw/providers/provider_manager.py
+++ b/src/copaw/providers/provider_manager.py
@@ -19,6 +19,7 @@
     ProviderInfo,
 )
 from copaw.providers.openai_provider import OpenAIProvider
+from copaw.providers.lm_studio_provider import LMStudioProvider
 from copaw.providers.anthropic_provider import AnthropicProvider
 from copaw.providers.ollama_provider import OllamaProvider
 from copaw.constant import SECRET_DIR
@@ -159,6 +160,15 @@
     require_api_key=False,
 )
 
+PROVIDER_LMSTUDIO = LMStudioProvider(
+    id="lmstudio",
+    name="LM Studio",
+    base_url="http://localhost:1234/v1",
+    require_api_key=False,
+    api_key_prefix="",
+    models=[],
+)
+
 
 class ModelSlotConfig(BaseModel):
     provider_id: str = Field(
@@ -216,6 +226,7 @@ def _init_builtins(self):
         self._add_builtin(PROVIDER_AZURE_OPENAI)
         self._add_builtin(PROVIDER_ANTHROPIC)
         self._add_builtin(PROVIDER_OLLAMA)
+        self._add_builtin(PROVIDER_LMSTUDIO)
         self._add_builtin(PROVIDER_LLAMACPP)
         self._add_builtin(PROVIDER_MLX)
 
@@ -419,6 +430,8 @@ def _provider_from_data(self, data: Dict) -> Provider:
             return AnthropicProvider.model_validate(data)
         if provider_id == "ollama" or chat_model == "OllamaChatModel":
             return OllamaProvider.model_validate(data)
+        if provider_id == "lmstudio":
+            return LMStudioProvider.model_validate(data)
         if data.get("is_local", False):
             return DefaultProvider.model_validate(data)
         return OpenAIProvider.model_validate(data)
diff --git a/website/public/docs/config.en.md b/website/public/docs/config.en.md
index 7cde01e2c2..8c5c973a66 100644
--- a/website/public/docs/config.en.md
+++ b/website/public/docs/config.en.md
@@ -288,11 +288,17 @@ CoPaw needs an LLM provider to work. You can set it up in three ways:
 
 ### Built-in providers
 
-| Provider   | ID           | Default Base URL                                    | API Key Prefix |
-| ---------- | ------------ | --------------------------------------------------- | -------------- |
-| ModelScope | `modelscope` | `https://api-inference.modelscope.cn/v1`            | `ms`           |
-| DashScope  | `dashscope`  | `https://dashscope.aliyuncs.com/compatible-mode/v1` | `sk`           |
-| Custom     | `custom`     | _(you set it)_                                      | _(any)_        |
+| Provider           | ID                  | Default Base URL                                    | API Key Prefix |
+| ------------------ | ------------------- | --------------------------------------------------- | -------------- |
+| ModelScope         | `modelscope`        | `https://api-inference.modelscope.cn/v1`            | `ms`           |
+| DashScope          | `dashscope`         | `https://dashscope.aliyuncs.com/compatible-mode/v1` | `sk`           |
+| Aliyun Coding Plan | `aliyun-codingplan` | `https://coding.dashscope.aliyuncs.com/v1`          | `sk-sp`        |
+| OpenAI             | `openai`            | `https://api.openai.com/v1`                         | _(any)_        |
+| Azure OpenAI       | `azure-openai`      | _(you set it)_                                      | _(any)_        |
+| Anthropic          | `anthropic`         | `https://api.anthropic.com`                         | _(any)_        |
+| Ollama             | `ollama`            | `http://localhost:11434`                            | _(none)_       |
+| LM Studio          | `lmstudio`          | `http://localhost:1234/v1`                          | _(none)_       |
+| Custom             | `custom`            | _(you set it)_                                      | _(any)_        |
 
 For each provider you need to set:
 
diff --git a/website/public/docs/config.zh.md b/website/public/docs/config.zh.md
index 5dfd95ee48..82e56c4df0 100644
--- a/website/public/docs/config.zh.md
+++ b/website/public/docs/config.zh.md
@@ -278,11 +278,17 @@ CoPaw 需要 LLM 提供商才能运行。有三种设置方式：
 
 ### 内置提供商
 
-| 提供商             | ID           | 默认 Base URL                                       | API Key 前缀 |
-| ------------------ | ------------ | --------------------------------------------------- | ------------ |
-| ModelScope（魔搭） | `modelscope` | `https://api-inference.modelscope.cn/v1`            | `ms`         |
-| DashScope（灵积）  | `dashscope`  | `https://dashscope.aliyuncs.com/compatible-mode/v1` | `sk`         |
-| 自定义             | `custom`     | _（你自己填）_                                      | _（任意）_   |
+| 提供商                 | ID                  | 默认 Base URL                                       | API Key 前缀 |
+| ---------------------- | ------------------- | --------------------------------------------------- | ------------ |
+| ModelScope（魔搭）     | `modelscope`        | `https://api-inference.modelscope.cn/v1`            | `ms`         |
+| DashScope（灵积）      | `dashscope`         | `https://dashscope.aliyuncs.com/compatible-mode/v1` | `sk`         |
+| 阿里云百炼 Coding Plan | `aliyun-codingplan` | `https://coding.dashscope.aliyuncs.com/v1`          | `sk-sp`      |
+| OpenAI                 | `openai`            | `https://api.openai.com/v1`                         | _（任意）_   |
+| Azure OpenAI           | `azure-openai`      | _（你自己填）_                                      | _（任意）_   |
+| Anthropic              | `anthropic`         | `https://api.anthropic.com`                         | _（任意）_   |
+| Ollama                 | `ollama`            | `http://localhost:11434`                            | _（无需）_   |
+| LM Studio              | `lmstudio`          | `http://localhost:1234/v1`                          | _（无需）_   |
+| 自定义                 | `custom`            | _（你自己填）_                                      | _（任意）_   |
 
 每个提供商需要设置：
 
diff --git a/website/public/docs/console.en.md b/website/public/docs/console.en.md
index 72378662fb..5e5ebcaef4 100644
--- a/website/public/docs/console.en.md
+++ b/website/public/docs/console.en.md
@@ -327,6 +327,28 @@ automatically when models are added/removed via Ollama CLI or Console.
 > `copaw models ollama-list`, `copaw models ollama-remove`. See
 > [CLI](./cli#ollama-models).
 
+### LM Studio provider
+
+The LM Studio provider connects to the LM Studio desktop application's
+OpenAI-compatible local server to discover and use loaded models.
+
+**Prerequisites:**
+
+- Install LM Studio from [lmstudio.ai](https://lmstudio.ai)
+- Load a model and start the local server in LM Studio (default: `http://localhost:1234`)
+
+**Configure:**
+
+1. Click **Settings** on the LM Studio provider card.
+2. The default Base URL is `http://localhost:1234/v1`. Adjust if needed, then
+   click **Save**.
+3. Click **Manage Models** to see models loaded in LM Studio. You can also
+   manually add model IDs.
+4. Select **LM Studio** in the **Provider** dropdown and pick a model.
+
+> LM Studio does not require an API key by default. Models must be loaded
+> in LM Studio before they appear in CoPaw.
+
 ### Choose the active model
 
 1. In the **LLM Config** section, select a **Provider** from the dropdown
@@ -375,18 +397,18 @@ Select rows → click **Delete** in the toolbar → confirm.
 
 ## Quick Reference
 
-| Page                  | Sidebar path                     | What you can do                                      |
-| --------------------- | -------------------------------- | ---------------------------------------------------- |
-| Chat                  | Chat → Chat                      | Talk with CoPaw, manage sessions                     |
-| Channels              | Control → Channels               | Enable/disable channels, configure credentials       |
-| Sessions              | Control → Sessions               | Filter, rename, delete sessions                      |
-| Cron Jobs             | Control → Cron Jobs              | Create/edit/delete jobs, run immediately             |
-| Workspace             | Agent → Workspace                | Edit persona files, view memory, upload/download     |
-| Skills                | Agent → Skills                   | Enable/disable/create/delete skills                  |
-| MCP                   | Agent → MCP                      | Enable/disable/create/delete MCP clients             |
-| Runtime Config        | Agent → Runtime Config           | Modify runtime configuration                         |
-| Models                | Settings → Models                | Configure providers, manage local/Ollama, pick model |
-| Environment Variables | Settings → Environment Variables | Add/edit/delete environment variables                |
+| Page                  | Sidebar path                     | What you can do                                                |
+| --------------------- | -------------------------------- | -------------------------------------------------------------- |
+| Chat                  | Chat → Chat                      | Talk with CoPaw, manage sessions                               |
+| Channels              | Control → Channels               | Enable/disable channels, configure credentials                 |
+| Sessions              | Control → Sessions               | Filter, rename, delete sessions                                |
+| Cron Jobs             | Control → Cron Jobs              | Create/edit/delete jobs, run immediately                       |
+| Workspace             | Agent → Workspace                | Edit persona files, view memory, upload/download               |
+| Skills                | Agent → Skills                   | Enable/disable/create/delete skills                            |
+| MCP                   | Agent → MCP                      | Enable/disable/create/delete MCP clients                       |
+| Runtime Config        | Agent → Runtime Config           | Modify runtime configuration                                   |
+| Models                | Settings → Models                | Configure providers, manage local/Ollama/LM Studio, pick model |
+| Environment Variables | Settings → Environment Variables | Add/edit/delete environment variables                          |
 
 ---
 
diff --git a/website/public/docs/console.zh.md b/website/public/docs/console.zh.md
index c48bda9429..0dcd9d7932 100644
--- a/website/public/docs/console.zh.md
+++ b/website/public/docs/console.zh.md
@@ -309,6 +309,24 @@ Ollama 提供商集成本地 Ollama 守护进程，动态加载其中的模型
 
 > 也可以通过 CLI 管理 Ollama 模型：`copaw models ollama-pull`、`copaw models ollama-list`、`copaw models ollama-remove`。详见 [CLI](./cli#ollama-模型)。
 
+### LM Studio 提供商
+
+LM Studio 提供商连接 LM Studio 桌面应用内置的 OpenAI 兼容本地服务器，自动发现并使用已加载的模型。
+
+**前置条件：**
+
+- 从 [lmstudio.ai](https://lmstudio.ai) 安装 LM Studio
+- 在 LM Studio 中加载模型并启动本地服务器（默认地址：`http://localhost:1234`）
+
+**配置步骤：**
+
+1. 点击 LM Studio 提供商卡片的 **设置** 按钮。
+2. 默认 Base URL 为 `http://localhost:1234/v1`，如有需要可修改，点击 **保存**。
+3. 点击 **模型** 查看 LM Studio 中当前已加载的模型，也可手动添加模型 ID。
+4. 在 **提供商** 下拉菜单中选择 LM Studio，选择模型后点击 **保存**。
+
+> LM Studio 默认不需要 API Key。模型必须在 LM Studio 中加载后才会在 CoPaw 中显示。
+
 ### 选择活跃模型
 
 1. 在顶部**LLM配置**的**提供商**下拉菜单中选择一个提供商（只显示已授权或
@@ -354,18 +372,18 @@ Ollama 提供商集成本地 Ollama 守护进程，动态加载其中的模型
 
 ## 快速索引
 
-| 页面     | 侧边栏路径        | 你能做什么                                         |
-| -------- | ----------------- | -------------------------------------------------- |
-| 聊天     | 聊天 → 聊天       | 和 CoPaw 对话、管理会话                            |
-| 频道     | 控制 → 频道       | 启用/禁用频道、填入凭据                            |
-| 会话     | 控制 → 会话       | 筛选、重命名、删除会话                             |
-| 定时任务 | 控制 → 定时任务   | 创建/编辑/删除任务、立即执行                       |
-| 工作区   | 智能体 → 工作区   | 编辑人设文件、查看记忆、上传/下载                  |
-| 技能     | 智能体 → 技能     | 启用/禁用/创建/删除技能                            |
-| MCP      | 智能体 → MCP      | 启用/禁用/创建/删除MCP                             |
-| 运行配置 | 智能体 → 运行配置 | 修改运行配置                                       |
-| 模型     | 设置 → 模型       | 配置提供商 API Key、管理本地/Ollama 模型、选择模型 |
-| 环境变量 | 设置 → 环境变量   | 添加/编辑/删除环境变量                             |
+| 页面     | 侧边栏路径        | 你能做什么                                                   |
+| -------- | ----------------- | ------------------------------------------------------------ |
+| 聊天     | 聊天 → 聊天       | 和 CoPaw 对话、管理会话                                      |
+| 频道     | 控制 → 频道       | 启用/禁用频道、填入凭据                                      |
+| 会话     | 控制 → 会话       | 筛选、重命名、删除会话                                       |
+| 定时任务 | 控制 → 定时任务   | 创建/编辑/删除任务、立即执行                                 |
+| 工作区   | 智能体 → 工作区   | 编辑人设文件、查看记忆、上传/下载                            |
+| 技能     | 智能体 → 技能     | 启用/禁用/创建/删除技能                                      |
+| MCP      | 智能体 → MCP      | 启用/禁用/创建/删除MCP                                       |
+| 运行配置 | 智能体 → 运行配置 | 修改运行配置                                                 |
+| 模型     | 设置 → 模型       | 配置提供商 API Key、管理本地/Ollama/LM Studio 模型、选择模型 |
+| 环境变量 | 设置 → 环境变量   | 添加/编辑/删除环境变量                                       |
 
 ---
 
diff --git a/website/public/docs/models.en.md b/website/public/docs/models.en.md
index 971174d9dd..702167f01f 100644
--- a/website/public/docs/models.en.md
+++ b/website/public/docs/models.en.md
@@ -4,7 +4,7 @@ You need to configure a model before chatting with CoPaw. You can do this under
 
 ![Console models](https://img.alicdn.com/imgextra/i1/O1CN01zHAE1Z26w6jXl2xbr_!!6000000007725-2-tps-3802-1968.png)
 
-CoPaw supports multiple LLM providers: **cloud providers** (require API Key), **local providers** (llama.cpp / MLX), and **Ollama provider**, and you can add **custom providers**. This page explains how to configure each type.
+CoPaw supports multiple LLM providers: **cloud providers** (require API Key), **local providers** (llama.cpp / MLX), **Ollama provider**, **LM Studio provider**, and you can add **custom providers**. This page explains how to configure each type.
 
 ---
 
@@ -115,6 +115,31 @@ The Ollama provider uses the **Ollama daemon** installed on your machine. Models
 > If you see `Ollama SDK not installed. Install with: pip install 'copaw[ollama]'`, make sure Ollama is installed from ollama.com and you’ve run `pip install 'copaw[ollama]'` in CoPaw’s environment. To remove a model, click **Models** on the Ollama card, then the **trash icon** next to the model and confirm.
 >
 > ![delete](https://img.alicdn.com/imgextra/i1/O1CN01OvNNu21shXVzD14go_!!6000000005798-2-tps-3802-1968.png)
+>
+> **Docker users:** If CoPaw runs inside a Docker container, `localhost` refers to the container — not your host machine. Change the Ollama Base URL to `http://host.docker.internal:11434` (and add `--add-host=host.docker.internal:host-gateway` to your `docker run` command). See the [Docker section in the README](https://github.com/agentscope-ai/CoPaw#using-docker) for details.
+
+## LM Studio provider
+
+The LM Studio provider connects to the **LM Studio** desktop application's built-in OpenAI-compatible server. Models are managed in the LM Studio GUI; CoPaw discovers loaded models automatically via the `/v1/models` endpoint.
+
+**Prerequisites:**
+
+- Install LM Studio from [lmstudio.ai](https://lmstudio.ai).
+- In LM Studio, load a model and start the local server (default: `http://localhost:1234`).
+
+1. On the Models page you'll see the LM Studio provider card.
+
+2. Click **Settings** at the bottom right. The default Base URL is `http://localhost:1234/v1`. Adjust if you changed the port in LM Studio. Click **Save**.
+
+3. Click **Models** to view models currently loaded in LM Studio. You can also manually add a model ID if needed.
+
+4. In **LLM Configuration** at the top, select **LM Studio** in the **Provider** dropdown and your model in the **Model** dropdown, then click **Save**.
+
+> **Tip:** LM Studio does not require an API key by default. If you have enabled authentication in LM Studio, enter the key in the **API Key** field. Models must be loaded in LM Studio's GUI before they appear in CoPaw.
+>
+> **Important — Context Length:** LM Studio loads models with a small default context length (often 2048 or 4096 tokens). CoPaw's system prompt (AGENTS.md + SOUL.md + PROFILE.md) can easily exceed this limit, causing an error like _"The number of tokens to keep from the initial prompt is greater than the context length"_. To fix this, **unload the model in LM Studio and reload it with a larger context length** (16384 or above is recommended). You can do this in the LM Studio GUI (Model Settings → Context Length) or via the CLI: `lms unload --all && lms load <model> -c 16384`.
+>
+> **Docker users:** If CoPaw runs inside a Docker container, `localhost` refers to the container — not your host machine. Change the LM Studio Base URL to `http://host.docker.internal:1234/v1` (and add `--add-host=host.docker.internal:host-gateway` to your `docker run` command). See the [Docker section in the README](https://github.com/agentscope-ai/CoPaw#using-docker) for details.
 
 ## Add custom provider
 
diff --git a/website/public/docs/models.zh.md b/website/public/docs/models.zh.md
index c7780e1749..cfce57af68 100644
--- a/website/public/docs/models.zh.md
+++ b/website/public/docs/models.zh.md
@@ -4,7 +4,7 @@
 
 ![控制台模型](https://img.alicdn.com/imgextra/i4/O1CN01XnOPPQ1c99vox3I88_!!6000000003557-2-tps-3786-1980.png)
 
-CoPaw 支持多种 LLM 提供商：**云提供商**（需 API Key）、**本地提供商**（llama.cpp / MLX）和 **Ollama 提供商**，且支持添加自定义 **提供商**。本文介绍这几类提供商的配置方式。
+CoPaw 支持多种 LLM 提供商：**云提供商**（需 API Key）、**本地提供商**（llama.cpp / MLX）、**Ollama 提供商**、**LM Studio 提供商**，且支持添加自定义 **提供商**。本文介绍这几类提供商的配置方式。
 
 ---
 
@@ -114,8 +114,33 @@ Ollama 提供商对接本机安装的 **Ollama 守护进程**，使用其中的
 
 > 如果在过程中遇到 `Ollama SDK not installed. Install with: pip install 'copaw[ollama]'`的提示，请先确认是否已经在 ollama.com 下载 Ollama，并在 CoPaw所在虚拟环境中执行过 `pip install 'copaw[ollama]'`。如果想删除某个模型，点击 Ollama 卡片右下角的 **模型**，在模型列表中，点击想要删除的模型右侧的 **垃圾桶按钮**，二次确认后即可删除。
 >
+> **Docker 用户：** 如果 CoPaw 运行在 Docker 容器中，`localhost` 指向的是容器自身而非宿主机。请将 Ollama 的 Base URL 改为 `http://host.docker.internal:11434`（并在 `docker run` 命令中添加 `--add-host=host.docker.internal:host-gateway`）。详见 [README 的 Docker 章节](https://github.com/agentscope-ai/CoPaw#使用-docker)。
+>
 > ![delete](https://img.alicdn.com/imgextra/i2/O1CN01p2o85m1Ul9rkY87PS_!!6000000002557-2-tps-3802-1968.png)
 
+## LM Studio 提供商
+
+LM Studio 提供商连接 **LM Studio** 桌面应用内置的 OpenAI 兼容服务器。模型在 LM Studio 的图形界面中管理，CoPaw 通过 `/v1/models` 端点自动发现已加载的模型。
+
+**前置条件：**
+
+- 从 [lmstudio.ai](https://lmstudio.ai) 安装 LM Studio。
+- 在 LM Studio 中加载模型并启动本地服务器（默认地址：`http://localhost:1234`）。
+
+1. 在控制台的模型页面中，可以看到 LM Studio 提供商对应的卡片。
+
+2. 点击右下角 **设置**，默认 Base URL 为 `http://localhost:1234/v1`。如果你在 LM Studio 中修改了端口，请相应调整。点击 **保存**。
+
+3. 点击 **模型** 查看 LM Studio 中当前已加载的模型。如有需要，也可以手动添加模型 ID。
+
+4. 在上方的 **LLM 配置** 中，**提供商** 对应的下拉菜单中选择 LM Studio，**模型** 对应的下拉菜单中选择想使用的模型。点击 **保存**。
+
+> **提示：** LM Studio 默认不需要 API Key。如果你在 LM Studio 中启用了认证功能，请在 **API Key** 字段中填入对应的密钥。模型必须在 LM Studio 的图形界面中加载后才会在 CoPaw 中显示。
+>
+> **重要 — 上下文长度：** LM Studio 加载模型时默认的上下文长度较小（通常为 2048 或 4096 tokens）。CoPaw 的系统提示词（AGENTS.md + SOUL.md + PROFILE.md）可能会超过此限制，导致报错 _"The number of tokens to keep from the initial prompt is greater than the context length"_。解决方法：**在 LM Studio 中卸载模型，然后以更大的上下文长度重新加载**（建议 16384 及以上）。可以在 LM Studio 图形界面中调整（模型设置 → Context Length），也可以通过 CLI 操作：`lms unload --all && lms load <model> -c 16384`。
+>
+> **Docker 用户：** 如果 CoPaw 运行在 Docker 容器中，`localhost` 指向的是容器自身而非宿主机。请将 LM Studio 的 Base URL 改为 `http://host.docker.internal:1234/v1`（并在 `docker run` 命令中添加 `--add-host=host.docker.internal:host-gateway`）。详见 [README 的 Docker 章节](https://github.com/agentscope-ai/CoPaw#使用-docker)。
+
 ## 添加自定义提供商
 
 1. 在控制台的模型页面点击 **添加提供商**。