Cherry Studio Setup: Use Claude and GPT via a Custom API

Cherry Studio is a free desktop AI client for Windows, macOS and Linux. It gives you a chat window with assistants, a model picker, knowledge bases, MCP tools and side-by-side answers from several models. Cherry Studio has no models of its own: you connect providers with API keys, and every message is billed by that provider.

APIVAI is one such provider. One APIVAI key works with Claude and GPT models, in both the OpenAI format and the Anthropic format, billed per token below official list prices. That means a single provider entry in Cherry Studio can give you Claude Sonnet 4.6, Claude Opus 5.5 and GPT-5.5 in the same model list.

Cherry Studio offers two clean ways to connect APIVAI:

Pick the custom provider if you want Claude and GPT side by side with one key; this is what most people should do. Pick the Anthropic provider if you only use Claude and prefer the native format, for example for Cherry Studio features that need an Anthropic-protocol endpoint. You can set up both with the same key.

In current versions the + Add Provider dialog shows separate endpoint fields (OpenAI and Anthropic by default, more under More Settings). The Cherry Studio docs don't say how a model chooses between two endpoints in one provider, so this guide fills in only the OpenAI endpoint there and uses the built-in Anthropic provider for the native Claude route. Older versions ask for a Provider Type instead; choose OpenAI.

1. Sign up at apivai.com with your email.
2. Top up your balance from $10 by card, crypto, Alipay or WeChat Pay. There is no subscription and no free trial; you pay only for the tokens you use.
3. Open the Dashboard, create a key and copy it. You can give the key its own budget limit.

1. Click Settings (the gear icon in the left navigation bar).
2. Open the Model Services tab.
3. Click + Add Provider below the provider list.
4. Fill in the dialog as in the table below and click Add.

About /v1: Cherry Studio expects the base address and appends the version and path itself, so https://api.apivai.com becomes https://api.apivai.com/v1/chat/completions. Current versions also accept https://api.apivai.com/v1 without doubling it, but never paste the full /v1/chat/completions address. A trailing # tells Cherry Studio to use an address exactly as typed; you don't need it for APIVAI.

1. Open the APIVAI provider you just created in the list.
2. Click Get model list. Cherry Studio fetches the models your key can use; click the + next to each model you want. Only added models appear in the model pickers.
3. If you prefer to type them, click + and enter a model ID such as claude-sonnet-4-6, claude-haiku-4-5 or gpt-5.5.

The model list changes over time, so take IDs from Get model list or the pricing page rather than from old screenshots.

1. Click Test (named Check in some versions) and pick a model. A success message means the key and address are right.
2. Turn the provider on with the switch in the upper-right corner of the provider page. An enabled provider shows a green dot in the list.
3. Open a chat, choose the model from the model drop-down at the top and send a message.

In Model Services, open Anthropic, paste your APIVAI key into API Key and change API Address from https://api.anthropic.com to https://api.apivai.com. Do not add /v1; the address has the same shape as Anthropic's own. Then click Get model list, add the Claude models you want and switch the provider on. GPT models will not work here; they use the custom provider from Step 2. The Claude API proxy page describes the Anthropic format in more detail.

Prices per 1M tokens as of 2026-10-07. All models are on the pricing page.

A practical setup is a cheap default assistant on Claude Haiku 4.5 or GPT-6 Luna, a main assistant on Claude Sonnet 4.6, and Claude Opus 5.5 only for the questions that need it. Cherry Studio lets you send one question to several models at once, which is a cheap way to compare them on your own tasks.

Cherry Studio itself is free; you pay only for tokens. Each message resends the conversation so far, so long chats cost more per turn. Thinking is on by default and thinking tokens are billed as output.

• Everyday chat: assume a turn of about 3,000 input tokens (your message plus recent history) and 800 output tokens. On Claude Sonnet 4.6 one turn costs $0.0073 ($0.021 at official prices). At 40 turns a day for 30 days (1,200 turns) that is $8.76 a month, against $25.20. The same month costs $2.93 on Claude Haiku 4.5 and $0.126 on GPT-6 Luna.
• Working with a long document: you paste a report of about 30,000 tokens and ask 10 follow-up questions, each resending about 31,000 input tokens and returning 1,500 output tokens. On Claude Sonnet 4.6 the session costs $0.401 ($1.16 at official prices).

To keep costs down, start a new topic when the subject changes, and give Cherry Studio its own key with a budget limit. The Dashboard lists every request with its tokens and cost. For a wider price comparison, see Claude API pricing compared.

Chat works fully through APIVAI, but knowledge bases need an embedding model to index your files, and APIVAI does not provide embeddings. To use a knowledge base, configure a separate embedding provider for it in Cherry Studio; the assistant that answers can still be a Claude or GPT model from APIVAI. Image generation models are not available through APIVAI either.

• 401 or "invalid API key": the key was copied incompletely, has been deleted, or has reached its budget limit. Copy it again from the Dashboard and check your balance.
• 404: the address is wrong. For the custom provider use https://api.apivai.com, without /chat/completions and without a trailing #. For the Anthropic provider use https://api.apivai.com without /v1.
• Model not found: the model ID must match exactly, for example claude-sonnet-4-6, not a display name. Use Get model list to see what your key can access. GPT models fail on the Anthropic provider; add them to the custom provider.
• Model missing from the chat picker: the provider switch is off, or you fetched the list but didn't click + next to the model.
• Empty or cut-off replies: thinking tokens count towards the output limit. In the assistant's Edit Assistant → Model tab, either turn off Enable message length limit (MaxToken) or set it to 4096 or more.
• Replies arrive all at once or the stream stops: check the Streaming output (Stream) switch in the same tab, and any proxy setting in Cherry Studio's network options.
• 429: the default limit is 60 requests per minute per key. Comparing several models at once uses several requests; wait a moment or ask support for a higher limit.

The docs list all endpoints and error codes.

Use a custom provider with the OpenAI endpoint for Claude and GPT together. Use the built-in Anthropic provider only if you want Claude in the native Anthropic format.

No need. Enter https://api.apivai.com; Cherry Studio appends /v1 and the request path itself. Never enter the full /v1/chat/completions address.

Yes. One APIVAI key works for every model on the pricing page and in both formats. The same key also works in Chatbox and other OpenAI-compatible tools.

No. Request and response content is not logged; only usage data such as model, token counts and cost is kept for billing. Your chat history stays in Cherry Studio on your computer.

No. Billing is prepaid and pay as you go, from a $10 top-up, with no subscription.

Create an account, top up from $10 and paste your key into Cherry Studio.