> ## Documentation Index
> Fetch the complete documentation index at: https://lmm.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Grok Build

在 `~/.grok/config.toml` 里增加一个自定义模型。对话模型使用 OpenAI Chat Completions，Base URL 保留 `/v1`。

| 配置项 | 值 |
| - | - |
| Base URL | `https://api.lmm.best/v1` |
| API Key | 环境变量 `LMM_API_KEY` |
| 协议 | `api_backend = "chat_completions"` |

```toml theme={null}
[model.lmm]
model = "gpt-5.6-sol"
name = "LMM"
base_url = "https://api.lmm.best/v1"
env_key = "LMM_API_KEY"
```

<CodeGroup>
  ```bash macOS / Linux theme={null}
  export LMM_API_KEY=sk-xxxxxxxx
  ```

  ```powershell Windows PowerShell theme={null}
  $env:LMM_API_KEY="sk-xxxxxxxx"
  ```
</CodeGroup>

启动 Grok 后，在模型列表里选择 **LMM**。`model` 换成控制台 **模型与价格** 里可用的模型 ID。

调用 Claude 时把协议改成 Messages，并带上 Anthropic 版本头：

```toml theme={null}
[model.lmm-claude]
model = "claude-sonnet-4-6"
name = "LMM Claude"
base_url = "https://api.lmm.best/v1"
api_backend = "messages"
env_key = "LMM_API_KEY"
extra_headers = { "anthropic-version" = "2023-06-01" }
```