Home → Guides → Cherry Studio with a custom API: address, models, Claude and image input

Cherry Studio with a custom API: address, models, Claude and image input

Cherry Studio adds a provider in under a minute. The one field people get wrong is the API address, because Cherry adds /v1 on its own. And once you start attaching screenshots, it matters which models can actually see them — we tested seven.

Adding the provider

Open Settings → Model Provider and click Add. Give it a name (for example APICLAN) and choose the provider type OpenAI.
Paste your API key from the console.
Enter https://apiclan.us as the API address.
Click Check and pick any model to confirm the key works.

You enter	Cherry Studio calls
`https://apiclan.us`	`https://apiclan.us/v1/chat/completions` ✓
`https://apiclan.us/v1/`	`https://apiclan.us/v1/chat/completions` ✓
`https://apiclan.us/v1`	`https://apiclan.us/v1/v1/chat/completions` ✗ 404

Cherry Studio appends /v1/chat/completions to the address unless it ends with /. So https://apiclan.us and https://apiclan.us/v1/ both work, but https://apiclan.us/v1 without the trailing slash becomes /v1/v1 and returns 404.

Adding models

Click Manage under the provider. Cherry Studio fetches the list from /v1/models, which returns only the models of your key’s group. On 12 October 2026 that was:

GPT group: gpt-5.6-sol, gpt-5.6-terra, gpt-6-astra, gpt-6-sol, gpt-6.1-sol
DeepSeek group: deepseek-v4-flash, deepseek-v4-pro, deepseek-v4.1-flash (plus dated versions)

Click + next to the ones you want. If a model is missing from the list, your key belongs to a different group — every key on this gateway is tied to one group.

Adding Claude

Add a second provider with the type Anthropic, the same address https://apiclan.us, and a key created in a Claude group. Then add claude-opus-5-5 or claude-fable-5-1. One provider per group keeps the keys straight: GPT, DeepSeek and Claude each get their own entry.

Which models read the images you attach

We sent each model the same small image filled with a single colour through the OpenAI-style request Cherry Studio uses, and asked what colour it was. We repeated it with a second colour so a lucky guess could not pass.

Model	Reads images	Answer time
`claude-opus-5-5`	✓	3.1 s
`claude-fable-5-1`	✓	3.3 s
`deepseek-v4.1-flash`	✓	2.3–3.1 s
`gpt-6-astra`	✓	6.6 s
`gpt-5.6-sol`	✓	7.3 s
`gpt-6.1-sol`	✓	24–25 s
`deepseek-v4-pro`	✗	—

deepseek-v4-pro does not accept images: it answered “white” for a green image and returned nothing the other time, without an error. For screenshots, use a model with a tick. gpt-6.1-sol reads images correctly but took about 24 seconds, against 2.6 seconds for the first token of a text-only request. If Cherry Studio shows no image button for a model, open the model’s settings and tick Vision. Tested 12 October 2026.

Common errors

401 → the key is wrong, or the key and the provider do not match (a GPT key in the Anthropic provider, for example).
404 → the address ends in /v1 without a slash. Use https://apiclan.us.
model_not_found (is not supported by any configured account in this group) → the model is not in your key’s group. Add it from Manage rather than typing it.
The reply ignores your image → the model cannot read images. Switch to one with a tick in the table above.

Setting up coding tools? Cline with a custom API · Codex with a custom API.

Start using it

No subscription, no monthly minimum, no sales call. Top up with USDT and spend what you use — 1 USDT gives you 2 credits of API balance.

Read the 30-second quickstart

Language: English · 简体中文 · 繁體中文 · 日本語 · 한국어 · Deutsch · Français · Español · Türkçe · Português · Italiano · Bahasa Indonesia · Tiếng Việt · Bahasa Melayu · Polski · Čeština · فارسی · Русский

Prices on this page are filled in automatically from live billing data. Third-party terms are quoted from that party's own documentation.