Skip to content
nikcli/inference

Cursor

Cursor accepts any OpenAI-compatible endpoint. Configure once, use every model in the catalog including aliases and :thinking variants.

1. Get your API key

Dashboard → API Keys → Create key → copy.

nik_live_xxxxxxxxxxxxxxxxx

2. Configure Cursor

Settings → Cursor Settings → Models

SettingValue
Override OpenAI Base URLtoggle ON
OpenAI Base URLhttps://inference.nikcli-ai.dev/v1
OpenAI API Keyyour nik_live_…

3. Add model names

Scroll down to Model Names → + Add model — add the ones you want:

nikseek
nikcli-fast
nikcli-coder
nikcli-reason
nikseek:thinking
deepseek-v4-pro
kimi-k2.6:thinking
qwen-3.5-flash

Click Verify — Cursor pings the endpoint to confirm the key works.

4. Pick the model in the chat

Hit the model picker (⌘/) and type the model id. The full nikcli catalog will autocomplete.

Tips