Cherry Studio Configuration Guide

Last updated: October 17, 2026 (Asia/Shanghai)
Related: Help Center · LLM API

Cherry Studio is a desktop client that supports multiple model providers. It is a good fit if you do not want to write code but still want to switch between models such as Claude / Gemini / Kimi in a graphical interface. TokenExplore works well with New API / OpenAI-compatible protocols, so the setup takes only a few steps.

1. Prepare a token

Sign in to the TokenExplore console and open "Token Management".
Click "Add Token": the name is up to you (e.g. cherry); set the expiration as needed; keep the default group if you are unsure.
Set a reasonable quota cap first (we recommend a separate token for each client, rather than one key for everything).
After saving, copy the key that starts with sk-.

Warning

Never post your key in group chats or screenshots. If it leaks, disable and recreate it immediately.

2. Download and install

Go to the Cherry Studio website or GitHub Releases to download the installer for your system:

Windows: .exe installer
macOS: .dmg (choose the matching Intel / Apple Silicon build or the universal build)
Linux: .AppImage / .deb, etc.

3. Add a model provider

Open Cherry Studio and go to Settings → Model Providers in the lower-left corner.
Click "Add" at the bottom.
For the provider type, prefer New API; if that option is not available, choose OpenAI or OpenAI Compatible.
Fill in:
- API Address: https://ai.afp.net (usually do not add /v1; Cherry appends it automatically)
- API Key: your sk-xxxxxxxx
Click "Check" to test connectivity, then "Manage" to fetch models and select the ones you need.

If you want to use Claude's native protocol, add another provider of type Anthropic, with the same API address https://ai.afp.net (without /v1) and the same key.

4. Add models

If automatic sync fails, you can add model IDs manually (they must match the platform exactly):

Use case	Recommended model ID
Everyday coding / chat	`claude-sonnet-4-6`
Stronger reasoning	`claude-opus-4-8`, `claude-opus-5`
Fast and inexpensive	`claude-haiku-4-5`, `gemini-3.5-flash`, `deepseek-v4-flash`
Gemini Pro	`gemini-3.1-pro`
Kimi / GLM / Qwen	`kimi-k3`, `glm-5.3`, `qwen3.8-max`

For the full list, see the model quick reference in the Help Center or GET /v1/models.

Go back to the chat page, select the provider and model you just configured at the top, and start chatting. Streaming is enabled by default.

5. FAQ

Connection failed or 404?
Check whether the address is the bare https://ai.afp.net (not the full /v1/chat/completions); whether the key is complete; and try switching between New API / OpenAI / OpenAI Compatible once.

How do I use Claude / Gemini?
In most cases you only need to add the corresponding model ID under the New API / OpenAI provider; you do not need official Google / Anthropic accounts. Add a separate Anthropic provider only when you need Anthropic-native features.

Knowledge base / Embeddings?
Cherry supports local document parsing. If the platform has not yet listed embedding models, knowledge-base vectorization may be unavailable — refer to /v1/models and the console; you can also use a client-side local embedding solution for now.

Why can't the client see the balance?
Check your balance and usage in the TokenExplore console.

Can I configure multiple providers?
Yes. TokenExplore, official keys, and other gateways can coexist, and you can switch between them as needed in a conversation.

Back to: Help Center

Cherry Studio Configuration Guide ​

1. Prepare a token ​

2. Download and install ​

3. Add a model provider ​