# Cherry Studio

Use a key from a **GPT group** (see [Groups](https://docs.2api.dev/en/start/endpoints/#groups)).

1. Open Settings → Model Provider in Cherry Studio and click Add.

2. Any name works, for example `2api`. Set the provider type to **OpenAI**.

3. Fill in the fields as shown below:

   | Field | Value |
   | --- | --- |
   | API Key | `sk-your-key` |
   | API Host (API address) | `https://2api.dev` |

   Enter **only the root address** as the API Host. Cherry Studio adds `/v1/chat/completions` itself. The gray preview below the box should show `https://2api.dev/v1/chat/completions`.

4. Click Manage to add models from the list. If the list can't be loaded, click Add and enter the model name by hand (look it up with [`/v1/models`](https://docs.2api.dev/en/start/endpoints/#model-names)).

5. Click Check next to the API key and choose a model to test. When it passes, you can pick the model in a conversation.

:::caution[Two ways to get the address wrong]
- Writing `https://2api.dev/v1`: it becomes `/v1/v1/chat/completions` and returns 404.
- Ending the address with `#`: Cherry Studio then treats the address as the complete request URL and adds no path. The `https://2api.dev/v1#` from older guides requests the wrong address in new versions. Change it to `https://2api.dev`.
:::

:::note[Want to use Claude in Cherry Studio?]
The Claude group may only allow calls from the Claude Code client. If the check shows an error such as `only allows Claude Code clients`, the group can only be used in [Claude Code](https://docs.2api.dev/en/clients/claude-code/).
:::