Run Cursor on Mingles Router
Cursor's Chat and Agent can run on any OpenAI-compatible endpoint: paste a key, override the base URL, add the model by name.
Pick your model and paste your key — every command and value on this page updates live.
Endpoint: https://router.mingles.ai/v1 · key stays in your browser ·
no key entered? commands show <your-key>.
Configure Cursor
1. Open the model settings
Gear icon → Cursor Settings → Models.
2. Paste your key and override the base URL
In the API Keys section paste your key into OpenAI API Key, enable Override OpenAI Base URL and set it to the value below. Hit Verify — it fires a test request at /chat/completions.
3. Add the model by name
In the model list press + Add model and type {{model}} exactly (case-sensitive).
4. Pick it in Chat / Composer
Select the model in the picker and code. Long agent sessions are where the per-token price difference compounds.
Exact values
| Where | Cursor Settings → Models → API Keys |
| Override OpenAI Base URL | https://router.mingles.ai/v1 |
| OpenAI API Key | <your-key> |
| Custom model name | deepseek-ai/DeepSeek-V4-Flash-0731 |
Switch model
Swap models by adding another name to the list — same key, same URL. MiniMaxAI/MiniMax-M2.7 burns the budget at half the rate of Kimi.
Before you file a bug: read the limits
Reasoning models spend the output budget on internal thinking, output is capped at 8192 tokens, there is no KV cache, and there are no built-in web tools. Most “it broke on Mingles Router” reports are one of these. See Model limits & behavior →
Frequently asked
Verify fails — what do I check? +
The base URL must end in /v1 (no trailing slash after it), and the key must be active. Test with curl against https://router.mingles.ai/v1/models if unsure.
Does Tab autocomplete use my custom key? +
No — Cursor Tab runs on Cursor’s own models regardless of the key. The custom endpoint powers Chat, Composer and Agent, which is where the token volume (and the bill) actually is.
Do I still need Cursor Pro? +
For Tab and the built-in frontier models, yes. For chat/agent usage on your own key you pay only your endpoint — many people run the free Cursor plan with a $9 key.