Cursor

Cursor lets you bring your own OpenAI API key and override the OpenAI base URL. Point that override at https://api.abliteratedllm.dev/v1 and add our model ids as custom models.

Placeholders: $API_BASE is https://api.abliteratedllm.dev/v1, $API_ORIGIN is https://api.abliteratedllm.dev, and $YOUR_KEY is your API key.

Configure

1. Open Cursor Settings and go to Models.
2. In the OpenAI API Key field, paste $YOUR_KEY.
3. Turn on Override OpenAI Base URL and enter $API_BASE.
4. Save, then add a custom model for each of our model ids you want to use. Any id from GET /v1/models works, currently glm-5.3-flash-abliterated, deepseek-v4.1-flash or deepseek-v4-pro, and it must match exactly.
5. Enable the custom models and pick one in the Chat or Agent model picker.

Good to know

• Use a public URL. Cursor routes your requests through its own servers to build the final prompt, so the base URL must be reachable from the internet. A localhost address won't work.
• Chat and Agent only. Your own key applies to chat models. Tab completion keeps using Cursor's built-in models.
• Plan. Cursor's own-key support depends on your Cursor plan; check Cursor's pricing page if the API key fields are missing.
• Zero data retention. Cursor's own zero-data-retention policy doesn't apply to requests made with your own key. On our side, prompts and outputs aren't stored (see the privacy policy). Models labelled Official DeepSeek API are the exception on retention: DeepSeek stores what you send them. See Privacy.
• Text only. Images in a request fail with 400 unsupported_content.

Verified against Cursor docs on 2026-11-03: Bring your own API key and Models and pricing. Cursor's page documents the OpenAI key field under Cursor Settings > Models; the Override OpenAI Base URL toggle sits next to it in the app but isn't described on that page.