Choosing a call format and base URL
6090ti offers two call formats: OpenAI-compatible and Anthropic-compatible. One API key works with both and calls every model enabled for you. This article explains how to pick a format, which base URL to enter, and common mistakes.
The two call formats
| Format | Base URL | Typically for | Suitable tools |
|---|---|---|---|
| Anthropic-compatible | https://api.6090ti.com |
Claude models | Claude Code, Cline, Roo Code, Cherry Studio, Anthropic SDK |
| OpenAI-compatible | https://api.6090ti.com/v1 |
GPT and other models | Codex, OpenCode, OpenAI SDK |
These are recommended combinations. To see which formats a model supports, use the model list endpoint.
How to choose
- Decide which tool you will use and check which call format it supports.
- Create an API key in the console (just give it a name).
- Request
/v1/modelswith this key and confirm that the model is in the returned list. The Model pricing page shows prices. - Follow the Integration guides to enter the base URL, key and model name.
One key can go into several tools. To see each tool’s usage separately, create a key per tool or computer.
Common mistakes
| Symptom | Cause |
|---|---|
| 404, model not available | The model name is wrong, or the model is not enabled for your key. |
| 404 on the base URL | The address is missing /v1 or has it twice. The OpenAI format needs /v1; for the Anthropic format enter https://api.6090ti.com as the base URL and let the client append /v1/messages. |
How do I check which models a key can use?
Request the model list with the key. It returns the models enabled for that key:
curl https://api.6090ti.com/v1/models -H "Authorization: Bearer <your key>"