Integration
Cursor with your own OpenAI-compatible API key
Cursor ships with a built-in model list, but it also accepts your own OpenAI-compatible endpoint. Three fields are enough: base URL, key and model name.
Updated
Where the three fields go
Open Settings → Models and scroll to the OpenAI API Key section. Paste the LLMcode key, enable the override and set the base URL to https://llmcode.io/v1. Without the override Cursor keeps calling OpenAI with your LLMcode key and every request fails with 401.
Then add the model name. Cursor needs the exact published slug, for example claude-sonnet-5-5 or deepseek-flash. If a slug is not in the picker, add it manually as a custom model with the same name.
- Override base URL: https://llmcode.io/v1
- API key: sk-llmcode-… from the cabinet
- Custom model name: a published slug from /models
Why a separate key for the editor
Cursor is a desktop app, so the key lives on a developer machine. A dedicated key with a spend limit, an expiry date and an optional IP list limits the damage if a laptop is lost and lets you revoke editor access without breaking a server integration.
The same key can also carry a model allowlist. Cursor will then see only those profiles in the picker, which keeps autocomplete on a cheap profile while chat stays on a larger one.
When Cursor does not see the models
Check the endpoint first: GET https://llmcode.io/v1/models with the same key must return a list. An empty list means the key has no allowed models; a 401 means the key was revoked, expired or copied with a stray space.
Cursor caches the model list per session. After changing the key or adding a custom model, restart the app before testing again.
Connect
Settings → Models → OpenAI API Key
Override OpenAI Base URL: https://llmcode.io/v1
API Key: sk-llmcode-…
Custom model name: claude-sonnet-5-5curl https://llmcode.io/v1/models \
-H "Authorization: Bearer $LLMCODE_API_KEY"Published model prices
USD per 1M tokens. The numbers come from the public catalog on every page rebuild — this is the sale price, not an estimate.
| Model | Input | Output | Cache |
|---|---|---|---|
| Claude Opus 5.5 | $2.70 | $8.00 | $0.14 |
| DeepSeek Flash | $0.50 | $2.00 | $0.01 |
| Claude Sonnet 5.5 | $1.33 | $6.67 | $0.07 |
| DeepSeek V4 Pro | $2.20 | $6.60 | $0.07 |
Top-up from $10, balance in USD. 1 USD = 1,000,000 UT. No subscription and no auto-charge.
Questions
Does Cursor work with Claude models through LLMcode?
Yes, when the model is a published slug. Cursor sends OpenAI-format requests, and the LLMcode profiles answer in that format regardless of the upstream vendor.
What does Cursor cost per month?
There is no separate Cursor fee. You pay only for tokens at the published per-1M rates of the chosen profile, and the balance is prepaid from $10.
Why do I get 401 after pasting the key?
Most often the base URL override is off, so Cursor sends the request to OpenAI. Enable the override and save the settings again.
Why do I get 402 or 429?
402 means the balance or the key limit is exhausted. 429 means the request rate is too high; the default ceiling is 600 requests per minute.