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

# Subscription-based providers

> Use supported subscriptions such as ChatGPT, Claude, Gemini, Grok, GitHub Copilot, GLM Coding Plan, Kimi, and Mistral Vibe.

If you already pay for ChatGPT Plus, Claude Max, GitHub Copilot, or one of the other plans listed below, the gateway can route through the subscription instead of an API key. Auth is OAuth, a device code, or a pasted subscription token, depending on the provider.

## Supported subscriptions

| Provider                                              | Plan                 | Auth flow          |
| ----------------------------------------------------- | -------------------- | ------------------ |
| [OpenAI](https://openai.com/chatgpt)                  | ChatGPT Plus/Pro     | OAuth (browser)    |
| [Anthropic](https://www.anthropic.com/claude)         | Claude Pro/Max       | Subscription token |
| [Google](https://gemini.google.com)                   | Gemini               | OAuth (browser)    |
| [xAI](https://x.ai)                                   | Grok                 | OAuth (browser)    |
| [GitHub Copilot](https://github.com/features/copilot) | Copilot              | Device code        |
| [MiniMax](https://www.minimaxi.com)                   | Coding Plan          | Device code        |
| [Kiro](https://kiro.dev)                              | Kiro                 | Device code        |
| [Z.ai (Zhipu)](https://z.ai)                          | GLM Coding Plan      | Subscription token |
| [Alibaba Cloud (Qwen)](https://qwen.ai)               | Qwen Token Plan      | Subscription token |
| [Moonshot](https://platform.moonshot.cn)              | Kimi Coding Plan     | Subscription token |
| [Mistral](https://mistral.ai)                         | Vibe                 | Subscription token |
| [BytePlus](https://www.byteplus.com)                  | ModelArk Coding Plan | Subscription token |
| Xiaomi MiMo                                           | Token Plan           | Subscription token |
| [NousResearch](https://nousresearch.com)              | Subscription         | Subscription token |
| Command Code                                          | Subscription         | Subscription token |
| [ClinePass](https://cline.bot)                        | ClinePass            | Subscription token |
| [OpenCode](https://opencode.ai)                       | OpenCode Go          | Subscription token |
| [Ollama Cloud](https://ollama.com)                    | Ollama Cloud         | Subscription token |

<Note>
  The Kimi Coding Plan serves `k3`, `k3-256k`, `kimi-for-coding` and
  `kimi-for-coding-highspeed`. If a tier or a fallback still points at
  `kimi-k3`, click **Refresh models** on the connection and pin one of these
  ids instead.
</Note>

## Connect a subscription

<Steps>
  <Step title="Open the Subscriptions page">
    In the dashboard sidebar, open **Providers → Subscriptions** and click
    **Connect** on the provider.
  </Step>

  <Step title="Complete the auth flow">
    The gateway opens the provider's OAuth page, shows you a device code to enter,
    or asks you to paste a token, depending on the provider. Once it's saved,
    you're routed.
  </Step>
</Steps>

A model your plan just gained doesn't show up on its own. Click **Refresh models** on the connection to pick it up.

## How auth works per provider

<Tabs>
  <Tab title="ChatGPT">
    Standard OAuth against `auth.openai.com`. Once approved, requests route
    through the Codex backend at `chatgpt.com/backend-api/codex/responses`
    rather than the public OpenAI API. Tokens refresh automatically.
  </Tab>

  <Tab title="Claude">
    Generate a token with the Claude CLI (`claude setup-token`) and paste it into
    the gateway. Requests carry the `anthropic-beta: oauth-2025-04-20` header so
    Anthropic recognizes the subscription token. Available models match what your
    plan grants in claude.ai.
  </Tab>

  <Tab title="GitHub Copilot">
    GitHub's device-code flow. The gateway gives you a short user code, you visit
    [github.com/login/device](https://github.com/login/device), paste it, and
    approve. No API key is ever issued. Requests then go to
    `api.githubcopilot.com/chat/completions` using a short-lived token the gateway
    refreshes for you.
  </Tab>

  <Tab title="MiniMax">
    MiniMax's device-code flow. The gateway gives you a short user code to approve,
    then requests use the Anthropic protocol via `api.minimax.io/anthropic`
    (international) or `api.minimaxi.com/anthropic` (China).
  </Tab>

  <Tab title="GLM Coding Plan">
    Paste the subscription token issued by Z.ai and pick a region in the
    connect form. Requests route to `api.z.ai` (outside China, the default)
    or `open.bigmodel.cn` (China Mainland).
  </Tab>

  <Tab title="OpenCode Go">
    Paste the subscription token from your OpenCode account. Requests use the
    Anthropic protocol against `opencode.ai/zen/go`, authenticated via the
    `x-api-key` header.
  </Tab>
</Tabs>

<Note>
  Subscription auth is rate-limited by the provider's plan, not by the gateway. If
  you hit a plan ceiling, requests fall through to your next
  [fallback](/llm-gateway/docs/llm-gateway/docs/llm-gateway#fallback) model, usually a different provider on an API key.
</Note>

## Why mix subscriptions and API keys

A common setup: subscription as the primary (predictable monthly cost), API-key provider as the fallback for when you hit the plan limit or want a model the subscription doesn't include.

Pin your subscription model to your default or a custom tier and add API-key models to the [fallback list](/llm-gateway/docs/llm-gateway/docs/llm-gateway#configuration). The gateway handles the switch.