Cursor

Cursor can send its OpenAI requests to any OpenAI-compatible endpoint by overriding the OpenAI base URL and supplying your own key.

Setup

Open Settings → Models (also labeled Models & API Keys).
Under API Keys, expand OpenAI API Key and paste your UberLLM key (ull_...).
Enable Override OpenAI Base URL and set it to:
```
https://api.uberllm.dev/v1
```
Add a custom model by name that exists in the UberLLM catalog — for example deepseek/deepseek-v3.1 — in the model list.
Click Verify. Cursor sends a test request to confirm the key and base URL resolve.

Important behavior

Overriding the OpenAI base URL reroutes Cursor's OpenAI model requests to UberLLM. Add the UberLLM model(s) you want by name and select one of them; Cursor's built-in OpenAI model names (e.g. gpt-*) will also be sent to your override URL, so use a model id that UberLLM actually serves.
Some of Cursor's newest first-party "Composer"/agent models do not support custom API keys and will ignore the override. Use the standard model selector with your added custom model for BYO-endpoint chat/edit.
If Verify does nothing, confirm the base URL ends in /v1 and that the model name you added is present in GET https://api.uberllm.dev/v1/models.

Verify from the terminal

If Cursor's built-in verify is unclear, confirm the endpoint independently:

curl https://api.uberllm.dev/v1/chat/completions \
  -H "Authorization: Bearer $UBERLLM_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"deepseek/deepseek-v3.1","messages":[{"role":"user","content":"ping"}]}'

A 200 with a choices[0].message body means Cursor will work with the same settings.