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

# GitHub Copilot | AI Gateway

> Route GitHub Copilot Chat through the Orq.ai AI Gateway using its Bring Your Own Key (BYOK) Custom Endpoint provider. Reach the full model catalog with unified tracing and cost tracking.

[GitHub Copilot Chat](https://code.visualstudio.com/docs/agent-customization/language-models) in VS Code supports [Bring Your Own Key (BYOK)](https://code.visualstudio.com/docs/agent-customization/language-models#_bring-your-own-language-model-key) through a **Custom Endpoint** provider. Pointing it at the **AI Gateway** routes Chat requests through **Orq.ai**, reaching the full model catalog with unified tracing, cost tracking, and access controls.

<Warning>
  Always use the `provider/model_id` format for the model `id` value (e.g. `anthropic/claude-sonnet-4-6`, not `claude-sonnet-4-6`). Bare IDs return a `400 invalid model format` error.
</Warning>

## Prerequisites

* VS Code 1.122 or later (the Custom Endpoint provider graduated from Insiders to stable in this release) with the GitHub Copilot Chat extension installed
* Active **Orq.ai** account
* [Orq.ai API key](/docs/ai-studio/organization/api-keys)
* Copilot Business or Enterprise orgs: an administrator must enable **Bring Your Own Language Model Key in VS Code** in the Copilot policy settings on GitHub.com

<Note>
  BYOK covers Chat and tool calls only. Inline suggestions, next-edit suggestions, and semantic search still require GitHub sign-in and are not routed through **Orq.ai**.
</Note>

## Setup

<Steps>
  <Step title="Open Manage Language Models">
    Run **Chat: Manage Language Models** from the Command Palette, or select the gear icon labeled **Manage Language Models** in the Chat model picker.
  </Step>

  <Step title="Add a Custom Endpoint provider">
    Select **Add Models**, then choose **Custom Endpoint** from the provider list. VS Code prompts for three things in sequence:

    * **Group name**: any label, e.g. `Orq.ai`.
    * **API key**: the **Orq.ai** [API key](/docs/ai-studio/organization/api-keys).
    * **API Type**: **Responses**, **Chat Completions**, or **Messages**, matching one of the tabs below.

    VS Code then opens the global `chatLanguageModels.json` file, in the VS Code user data directory, not the project workspace, with the group already scaffolded. The API key is stored as a secret reference (`"apiKey": "${input:chat.lm.secret...}"`), not the raw key; leave that field as generated.
  </Step>

  <Step title="Add a model to the group">
    `chatLanguageModels.json` is an array of groups. Add an entry to the scaffolded group's `models` array:

    <Tabs>
      <Tab title="Responses (recommended)" icon="https://mintcdn.com/orqai/d-t0Z04KwFlGVsS1/images/logos/openai.svg?fit=max&auto=format&n=d-t0Z04KwFlGVsS1&q=85&s=41cb629a8b2739dc90fac2ddea651c17" width="395" height="400" data-path="images/logos/openai.svg">
        Matches the API Type selected as **Responses**. Reaches the full **AI Gateway** catalog through **Orq.ai**'s primary API.

        ```json theme={"theme":{"light":"github-light","dark":"github-dark"}}
        [
          {
            "name": "Orq.ai",
            "vendor": "customendpoint",
            "apiKey": "${input:chat.lm.secret...}",
            "apiType": "responses",
            "models": [
              {
                "id": "openai/gpt-5.4",
                "name": "GPT-5.4 via Orq",
                "url": "https://api.orq.ai/v3/router/responses",
                "toolCalling": true,
                "vision": true,
                "maxInputTokens": 1050000,
                "maxOutputTokens": 128000
              }
            ]
          }
        ]
        ```

        Replace `openai/gpt-5.4` with any identifier from the **AI Gateway** catalog (e.g. `anthropic/claude-sonnet-4-6`, `google/gemini-2.0-flash-001`), and adjust `maxInputTokens`/`maxOutputTokens` to match the chosen model.
      </Tab>

      <Tab title="Chat Completions" icon="https://mintcdn.com/orqai/d-t0Z04KwFlGVsS1/images/logos/openai.svg?fit=max&auto=format&n=d-t0Z04KwFlGVsS1&q=85&s=41cb629a8b2739dc90fac2ddea651c17" width="395" height="400" data-path="images/logos/openai.svg">
        Matches the API Type selected as **Chat Completions**. Reaches the full **AI Gateway** catalog through the OpenAI-compatible router.

        ```json theme={"theme":{"light":"github-light","dark":"github-dark"}}
        [
          {
            "name": "Orq.ai",
            "vendor": "customendpoint",
            "apiKey": "${input:chat.lm.secret...}",
            "apiType": "chat-completions",
            "models": [
              {
                "id": "openai/gpt-5.4",
                "name": "GPT-5.4 via Orq",
                "url": "https://api.orq.ai/v3/router/chat/completions",
                "toolCalling": true,
                "vision": true,
                "maxInputTokens": 1050000,
                "maxOutputTokens": 128000
              }
            ]
          }
        ]
        ```
      </Tab>

      <Tab title="Messages" icon="https://mintcdn.com/orqai/d-t0Z04KwFlGVsS1/images/logos/anthropic.svg?fit=max&auto=format&n=d-t0Z04KwFlGVsS1&q=85&s=b097662b141ad6f89f0c8d039ae241dc" width="61" height="43" data-path="images/logos/anthropic.svg">
        Matches the API Type selected as **Messages**. Routes Claude models through the Anthropic-compatible endpoint.

        ```json theme={"theme":{"light":"github-light","dark":"github-dark"}}
        [
          {
            "name": "Orq.ai Anthropic",
            "vendor": "customendpoint",
            "apiKey": "${input:chat.lm.secret...}",
            "apiType": "messages",
            "models": [
              {
                "id": "anthropic/claude-sonnet-4-6",
                "name": "Claude Sonnet 4.6 via Orq",
                "url": "https://api.orq.ai/v3/anthropic/v1/messages",
                "toolCalling": true,
                "vision": true,
                "maxInputTokens": 200000,
                "maxOutputTokens": 64000
              }
            ]
          }
        ]
        ```

        Replace `anthropic/claude-sonnet-4-6` with any Claude identifier from the **AI Gateway** catalog.
      </Tab>
    </Tabs>
  </Step>

  <Step title="Select the model in Copilot Chat">
    Open the Chat model picker and select a model from the new group to start using it.

    <Frame caption="The configured model appears in the Chat model picker under its group name.">
      <img src="https://mintcdn.com/orqai/ifpzhO2hwITYtXos/images/copilot-select-model.png?fit=max&auto=format&n=ifpzhO2hwITYtXos&q=85&s=3e547d0a446d4233d40be08efcba2bdb" alt="Chat model picker showing GPT-5.4 via Orq selected under the Orq.ai group, alongside built-in models and a Manage Models option." width="646" height="466" data-path="images/copilot-select-model.png" />
    </Frame>
  </Step>
</Steps>

## Configuration Reference

`chatLanguageModels.json` is an array of provider groups.

| Field                                | Level | Description                                                                                                                                                                                         |
| ------------------------------------ | ----- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `vendor`                             | Group | Must be `"customendpoint"` to enable BYOK.                                                                                                                                                          |
| `apiKey`                             | Group | Set automatically to a secret reference when the **Orq.ai** [API key](/docs/ai-studio/organization/api-keys) is entered in the Add Models prompt. Do not paste the raw key here.                    |
| `apiType`                            | Group | `"responses"` for **Orq.ai**'s primary API, `"chat-completions"` for the OpenAI-compatible router, or `"messages"` for the Anthropic endpoint.                                                      |
| `id`                                 | Model | **AI Gateway** model identifier in `provider/model_id` format.                                                                                                                                      |
| `url`                                | Model | Full endpoint URL, not a base URL: `https://api.orq.ai/v3/router/responses`, `https://api.orq.ai/v3/router/chat/completions`, or `https://api.orq.ai/v3/anthropic/v1/messages`, matching `apiType`. |
| `toolCalling`                        | Model | Set to `true` for models that support function calling.                                                                                                                                             |
| `vision`                             | Model | Set to `true` for models that accept image input.                                                                                                                                                   |
| `maxInputTokens` / `maxOutputTokens` | Model | Context window limits for the selected model.                                                                                                                                                       |

## Troubleshooting

<AccordionGroup>
  <Accordion title="Requests failing with 401 Unauthorized">
    The secret behind `apiKey` is likely wrong or expired. Reopen **Manage Language Models**, remove the group, and re-add it, pasting a valid **Orq.ai** [API key](/docs/ai-studio/organization/api-keys) (not a GitHub, OpenAI, or Anthropic key) at the prompt.
  </Accordion>

  <Accordion title="400 invalid model format">
    Set the model `id` to the `provider/model_id` format (e.g. `anthropic/claude-sonnet-4-6`). Bare model IDs are rejected by the router.
  </Accordion>

  <Accordion title="Custom Endpoint model does not appear in the picker">
    Confirm `url` points to a full endpoint path, not just a base URL. `https://api.orq.ai/v3/router` alone is rejected; use `https://api.orq.ai/v3/router/responses` or `https://api.orq.ai/v3/router/chat/completions`.
  </Accordion>

  <Accordion title="No Traces appearing in Orq.ai">
    Confirm `url` matches `apiType`: `responses` must point to `https://api.orq.ai/v3/router/responses`, `chat-completions` must point to `https://api.orq.ai/v3/router/chat/completions`, `messages` must point to `https://api.orq.ai/v3/anthropic/v1/messages`. Requests sent to the wrong endpoint will not appear in [**Orq.ai** Traces](/docs/ai-studio/observability/traces).
  </Accordion>

  <Accordion title="Inline suggestions still use a different model">
    BYOK only applies to Chat and tool calls. Inline suggestions, next-edit suggestions, and semantic search continue to require GitHub sign-in and are not affected by the Custom Endpoint configuration.
  </Accordion>

  <Accordion title="Command Palette opens 'Manage Language Model Access' instead of the Custom Endpoint editor">
    On a Copilot Business or Enterprise account, **Manage Language Models** does not exist as a command until an administrator enables **Bring Your Own Language Model Key in VS Code** in the Copilot policy settings on GitHub.com. The Command Palette falls back to the similarly named **Manage Language Model Access** command instead, which controls extension permissions and is unrelated to BYOK. Ask a workspace administrator to enable the policy, then reload VS Code.
  </Accordion>

  <Accordion title="Custom Endpoint is missing from the Add Models list">
    The Custom Endpoint provider requires VS Code 1.122 or later; earlier versions only show built-in providers (Anthropic, Azure, Google, Ollama, OpenAI, OpenRouter, xAI). Update VS Code via **Code → Check for Updates** (macOS) or **Help → Check for Updates**, then reopen **Manage Language Models**.
  </Accordion>

  <Accordion title="'No utility model is configured' error when sending a message">
    Expected the first time a BYOK model is used as the active main model. Open Settings (<kbd>Cmd+,</kbd>), search **utility**, and set **Chat: Byok Utility Model Default** to one of the configured Custom Endpoint models.
  </Accordion>
</AccordionGroup>

## Verification

Select the configured model in the Chat model picker and send a message. The response appears in Copilot Chat and the trace appears in [**Orq.ai** Traces](/docs/ai-studio/observability/traces) with the model identifier set in `id`.