# Configuring Copilot in GitHub Desktop Choose which AI model GitHub Desktop 中的Copilot uses for each feature, or connect your own LLM provider. You can choose which model Copilot uses for each GitHub Desktop feature, such as commit message generation and conflict resolution. You can also configure GitHub Desktop 中的Copilot to use your own LLM provider (BYOK) instead of GitHub-hosted models. This lets you connect to OpenAI-compatible endpoints, Azure OpenAI, or Anthropic, including locally running models such as Ollama. ## Prerequisites * You must be signed in to a GitHub account with access to GitHub Desktop 中的Copilot. * If your access is managed by an organization or enterprise, GitHub Desktop 中的Copilot must be enabled for your account. * You have an API key or bearer token from a supported LLM provider, or you have a local model running, such as Ollama. * You have the base URL and at least one model identifier for the provider you want to use. > \[!NOTE] > Custom LLM providers in GitHub Desktop require access to GitHub Desktop 中的Copilot. To use your own LLM models with Copilot 命令行界面（CLI）, see [在 GitHub Copilot CLI 中使用自己的 LLM 模型](/zh/copilot/how-tos/copilot-cli/customize-copilot/use-byok-models). ## Supported provider types GitHub Desktop supports three custom provider types: | Provider type | Compatible services | | ------------------------------ | ------------------------------------------------------------------------------------------------------------ | | **OpenAI / OpenAI-compatible** | OpenAI, Ollama, vLLM, Foundry Local, and any other endpoint that is compatible with the selected API format. | | **Azure** | Azure OpenAI Service. | | **Anthropic** | Anthropic Claude models. | ## Configuring your provider You configure your model provider by adding a custom provider in GitHub Desktop settings.
1. 在菜单栏中，选择“GitHub Desktop”，然后单击“设置”\*\*\*\*\*\*\*\*。 ![Mac 上的菜单栏的屏幕截图。 在打开的“GitHub Desktop”下拉菜单下，光标悬停在“设置”上，该设置以蓝色突出显示。](/assets/images/help/desktop/mac-choose-settings.png) 2. In the left sidebar, click ** Copilot**. 3. Click the **Providers** tab. 4. Click **Add Provider**. 5. Under **Name**, type a name for the LLM provider. 6. Under **Type**, select the provider type. 7. Under **Base URL**, type the base URL of your model provider's API endpoint. The base URL must be an HTTPS URL, or an HTTP URL that points to the local machine. 8. If you selected **OpenAI / OpenAI-compatible**, under **API Format**, select the API format your provider expects. * Select **Chat completions (default)** for providers that use the OpenAI Chat Completions API. * Select **Responses (GPT-5 series)** for providers that use the OpenAI Responses API. 9. If you selected **Azure**, under **Azure API Version**, type the API version for your deployment. 10. Optionally, under **Request Timeout (seconds)**, type the number of seconds GitHub Desktop waits for the provider to respond. 11. Under **Authentication**, select the authentication method. * Select **API key** to authenticate with an API key. * Select **Bearer token** to authenticate with a bearer token. * Select **None** only for endpoints that do not require credentials. 12. If you selected **API key** or **Bearer token**, type the required credential. 13. Add at least one model to the provider. For more information, see [Adding models to your provider](#adding-models-to-your-provider). 14. Click **Add**.
1. 使用文件”菜单，然后单击“选项” 。 ![Windows 上的“GitHub Desktop”菜单栏的屏幕截图。 在展开的“文件”下拉菜单中，以橙色框出了“选项”项。](/assets/images/help/desktop/windows-choose-options.png) 2. In the left sidebar, click ** Copilot**. 3. Click the **Providers** tab. 4. Click **Add Provider**. 5. Under **Name**, type a name for the LLM provider. 6. Under **Type**, select the provider type. 7. Under **Base URL**, type the base URL of your model provider's API endpoint. The base URL must be an HTTPS URL, or an HTTP URL that points to the local machine. 8. If you selected **OpenAI / OpenAI-compatible**, under **API Format**, select the API format your provider expects. * Select **Chat completions (default)** for providers that use the OpenAI Chat Completions API. * Select **Responses (GPT-5 series)** for providers that use the OpenAI Responses API. 9. If you selected **Azure**, under **Azure API Version**, type the API version for your deployment. 10. Optionally, under **Request Timeout (seconds)**, type the number of seconds GitHub Desktop waits for the provider to respond. 11. Under **Authentication**, select the authentication method. * Select **API key** to authenticate with an API key. * Select **Bearer token** to authenticate with a bearer token. * Select **None** only for endpoints that do not require credentials. 12. If you selected **API key** or **Bearer token**, type the required credential. 13. Add at least one model to the provider. For more information, see [Adding models to your provider](#adding-models-to-your-provider). 14. Click **Add**.
## Adding models to your provider Add the models you want to use from your provider. Each model you add appears in the model picker alongside GitHub-hosted models. 1. In the **Add Custom Provider** or **Edit Custom Provider** dialog, under **Models**, click **Add Model**. 2. Under **Display Name**, type the friendly name shown in the Copilot model picker. 3. Under **Model Identifier**, type the exact model name your provider expects. 4. Under **Reasoning Effort**, select the reasoning level for the model. For non-reasoning models, or to let the provider choose, leave **Default (provider's choice)** selected. 5. Click **Add**. ## Reviewing provider responsibilities When you use your own LLM provider, GitHub Desktop sends prompts and repository context to that provider instead of GitHub. Check your provider's data handling and retention policies before adding credentials. Always review generated commit messages and conflict-resolution suggestions before committing. For more information, see [负责任地使用GitHub Copilot功能](/zh/copilot/responsible-use). ## Further reading * [在 GitHub Desktop 中提交并审查对项目的更改](/zh/desktop/making-changes-in-a-branch/committing-and-reviewing-changes-to-your-project-in-github-desktop) * [GitHub Copilot中支持的 AI 模型](/zh/copilot/reference/ai-models/supported-models) * [在 GitHub Copilot CLI 中使用自己的 LLM 模型](/zh/copilot/how-tos/copilot-cli/customize-copilot/use-byok-models)