Client Integration Overview
Don't want to write code? Most mainstream AI clients support a "custom OpenAI API endpoint". Fill in two things and you can use all the platform's models:
- API base URL:
https://gateway.mindproxy.ai/v1 - API key: the
sk-...you created in the console
General rule
The platform exposes an OpenAI-compatible API. Any client that supports a custom OpenAI API address works the same way: set the address to /v1, the key to your sk-..., and the model to a supported model ID. Per-client specifics are below.
Per-client guides
| Client | Platform | Good for |
|---|---|---|
| Cherry Studio | Win / Mac / Linux desktop | Full-featured, multi-model, knowledge base |
| ChatBox | Desktop / mobile / web | Simple to start |
| NextChat | Web / desktop | Self-hostable, lightweight |
| Cursor | Code editor | AI coding |
| Open WebUI | Self-hosted web | Team sharing, on-prem |
Does the base URL need /v1?
Clients handle the address differently — this is the most common gotcha:
- Most clients want the "API address / API host" set to
https://gateway.mindproxy.ai/v1. - A few clients append
/v1automatically, so you only fillhttps://gateway.mindproxy.ai. - If a call returns
404or "path not found",/v1is usually one too many or one too few — adjust it.
What model name to use
Use a public model ID the platform supports, e.g. gpt-4o-mini, claude-3-5-sonnet, gemini-1.5-pro.
- Full list: Models & plans, or the console "Model Prices" page.
- Or list via API:
GET /v1/models.
Common troubleshooting
| Symptom | Likely cause |
|---|---|
| 401 Unauthorized | Wrong key / extra whitespace / key disabled |
| 404 Not found | /v1 is one too many or too few in the base URL |
| Model unavailable / 403 | Model not in the key's scope, or model ID typo |
| 402 | Out of balance or over the key's monthly limit — see Recharge |
| 429 | Rate limited, retry later |
More in Errors & rate limits.