Cursor — Free LLM API Config

Configure Cursor with any free OpenAI-compatible API in under a minute.

How It Works

Cursor supports custom API providers via Settings → Models → Add Model. You enter a model name, base URL, and API key — Cursor routes completions through that provider.

Step-by-Step

Get a free API key from one of the providers below
Open Cursor Settings → Models → scroll to "OpenAI API Key" section
Override the API key — paste your free provider's key into the "OpenAI API Key" field
Override the base URL — paste the provider's base URL into "OpenAI Base URL"
Add the model name — under "Custom Models", add the model ID (e.g., gpt-oss-120b)

Recommended Backends

Groq

Base URL:  https://api.groq.com/openai/v1
API Key:   gsk_your_groq_api_key
Models:    llama-3.3-70b-versatile, qwen-3-32b, deepseek-r1-distill-llama-70b

Get key: console.groq.com/keys — no credit card, instant.

OpenRouter

Base URL:  https://openrouter.ai/api/v1
API Key:   sk-or-v1-your_openrouter_key
Models:    openai/gpt-oss-120b:free, google/gemini-2.5-flash:free, qwen/qwen3-coder:free

Get key: openrouter.ai/keys — 35+ free models.

NVIDIA NIM

Base URL:  https://integrate.api.nvidia.com/v1
API Key:   nvapi-your_nvidia_key
Models:    meta/llama-3.3-70b-instruct, deepseek-ai/deepseek-r1

Get key: build.nvidia.com — phone verification required, no credit card.

Cloudflare Workers AI

Base URL:  https://api.cloudflare.com/client/v4/{account_id}/ai/v1
API Key:   your_cloudflare_token
Models:    @cf/meta/llama-3.1-8b-instruct, @cf/deepseek-ai/deepseek-r1-distill-qwen-32b

Get key: dash.cloudflare.com — 10K req/day free.

Caveats

Cursor's agent mode (multi-step tool use) works best with models that support tool calling. Stick to Llama 3.3 70B+ or DeepSeek R1 for agent features.
Some providers throttle free-tier requests — if you see "rate limit" errors, try a different provider.
The model name in Cursor must match the provider's model ID exactly.

More Configs

Visit freellm.net/config/cursor/ for the interactive config generator.