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

# Global AI Token Management: Create, Limit, and Secure

> Create API tokens in the Global AI console: set expiration, remaining quota, model restrictions, IP allowlists, and groups.

A token (also called an API key) is the credential for calling Global AI. Each token has its own scope and quota cap. Click **API keys** in the left sidebar, or go directly to [https://globalai.vip/keys](https://globalai.vip/keys).

The token list shows every token you have created, including name, status, quota used, quota remaining, and expiration.

<h2 id="create-a-token">
  Create a token
</h2>

<Steps>
  <Step title="Open the create dialog">
    On the **API keys** list, click **Create API key** in the top right.
  </Step>

  <Step title="Fill in basic info">
    Enter a token name. Naming by purpose ("Production", "Testing", "Cherry Studio") makes it easier to distinguish tokens in the logs later.
  </Step>

  <Step title="Choose a key group">
    **For the group, we recommend the "auto" group by default.** With auto, a single key can call every supported model on the platform. The system routes requests to the correct billing group based on the model ID you use, which is convenient and flexible.

    If your needs are more specific (for example, only using the GPT series), you can pick the matching "gpt" group. In that case the key is **limited** to GPT-related models and cannot access models from other vendors, which suits scenarios with strict model-source constraints.
  </Step>

  <Step title="Configure advanced options (optional)">
    Set the [advanced options](#advanced-options) below as needed. Any option left blank means no restriction, though the account's total quota still applies.
  </Step>

  <Step title="Submit and save the token key">
    After confirming the details, click **Submit**. The new key (typically formatted as `sk-xxxxxxxx`) appears in the key list. Click the **Copy** button next to the key to grab it for use.
  </Step>
</Steps>

<Warning>
  A token key has full API call permissions. Do not share it, and do not commit it to public repositories or client applications. Store it in a password manager or a team key-management system.
</Warning>

<h2 id="advanced-options">
  Advanced options
</h2>

| Option            | Description                                                                                      |
| ----------------- | ------------------------------------------------------------------------------------------------ |
| Expiration        | Set a validity period. Blank means no expiration.                                                |
| Remaining quota   | Cap the quota this token can consume. When exhausted, the token becomes invalid.                 |
| Unlimited quota   | When enabled, the token has no per-token quota (the account's total still applies).              |
| Model restriction | Restrict the token to specific models. Blank means no restriction.                               |
| IP allowlist      | Restrict which source IPs may use the token. Blank means no restriction.                         |
| Group             | Specify the channel group this token uses. Available models and multipliers can differ by group. |

<Tip>
  Splitting tokens by purpose is best practice: one for production, one for testing, one for external partners. If one leaks you only need to delete that token without affecting the others.
</Tip>

## Use a token

Send the token key as `Authorization: Bearer <token>` (OpenAI-compatible format) or the native vendor header (for example `x-api-key`) to the platform's API base URL. See [Call the API](/en/usage/api) for full examples.

To quickly verify a token works, chat with a model in the built-in [Playground](/en/usage/api#playground-online-testing) without writing code.

## Edit a token

In the token list, find the target token and click the **Edit** icon on the right. You can change the name, expiration, remaining quota, model restrictions, IP allowlist, and other settings. **The token key itself cannot be changed**; only deleted.

<h2 id="delete-a-token">
  Delete a token
</h2>

In the token list, find the target token and click the **Delete** icon on the right. After confirmation the token becomes invalid immediately and cannot be recovered. Applications using it return an authentication error at once.

<Note>
  If a token leaks, delete it and create a new one immediately. Review the logs before deletion to check for suspicious calls.
</Note>

## Next steps

<CardGroup cols={2}>
  <Card title="Call the API" icon="code" href="/en/usage/api">
    Use your token to make OpenAI, Claude, or Gemini format calls.
  </Card>

  <Card title="Client integrations" icon="plug" href="/en/clients/overview">
    Configure tokens in chat clients and Agent tools.
  </Card>
</CardGroup>


## Related topics

- [Global AI FAQ: Accounts, Tokens, Calls, and Billing](/en/faq.md)
- [Call Global AI: OpenAI, Claude, and Gemini Formats](/en/usage/api.md)
- [Welcome to Global AI](/en/index.md)
- [Global AI Client Integrations: Parameters and Selection](/en/clients/overview.md)
- [Global AI Quickstart: From Sign-Up to First API Call](/en/quickstart.md)