Cursor

One /v1 endpoint for every provider. Point your OpenAI client at a new base URL and key, bring your own provider keys, and pay 0% markup on tokens.

Use any model in the catalogue from Cursor’s chat and agent, on your own provider keys. Cursor sends these requests to an OpenAI-compatible endpoint you name, so it needs three things from you: the base URL, an inference key and the model ids.

Set it up

  • Create an inference key under Keys in the dashboard, and add your provider key under BYOK.
  • In Cursor, open Cursor Settings → Models.
  • Paste the inference key into OpenAI API Key.
  • Turn on Override OpenAI Base URL and enter https://autonomousrelay.com/v1.
  • Add each model you want under custom models, by its id, such as anthropic/claude-sonnet-5 or openai/gpt-5.5.
What to expect. Custom keys work for chat and agent; Tab completion keeps using Cursor’s own models. Cursor sends the requests from its own servers, which is why the base URL must be a public address like ours. While the override is on, Cursor’s own models may refuse to run with it, so turn it off to use them. Cursor documents the key itself but describes the override only in its forum, so the names of these settings may change between versions.