One API, 300+ models — including many free ones
OpenRouter — One API, 300+ models — including many free ones.
Free tier: "Free" tagged models cost zero — but rate-limited globally to 20 requests/min, 200/day. Add $10 credit to lift to 1000/day.
API is OpenAI-compatible — point your SDK at https://openrouter.ai/api/v1.
meta-llama/llama-3.3-70b-instruct:freedeepseek/deepseek-chat:freegoogle/gemini-2.0-flash-exp:freeqwen/qwen-2.5-72b-instruct:freePay-as-you-go via prepaid credits. Most non-free models priced at OpenAI/Anthropic equivalents with a small markup; some cheaper than direct.
BYOK + enterprise contracts available.
https://openrouter.ai/api/v1
OpenAI-compatible — works with the OpenAI SDK by overriding base_url.
curl https://openrouter.ai/api/v1/chat/completions \
-H "Authorization: Bearer $OPENROUTER_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "meta-llama/llama-3.3-70b-instruct:free",
"messages": [{"role": "user", "content": "Hello in 5 words"}]
}'
A/B testing models, multi-provider failover, building agents that pick model per task. The fastest way to try every frontier model with one key.
Latency-critical use cases — adds a routing hop. Use Groq / direct provider when ms matter.
"Free" tagged models cost zero — but rate-limited globally to 20 requests/min, 200/day. Add $10 credit to lift to 1000/day. No credit card is required to sign up.
Most commonly used: meta-llama/llama-3.3-70b-instruct:free, deepseek/deepseek-chat:free, google/gemini-2.0-flash-exp:free, qwen/qwen-2.5-72b-instruct:free. The full current list is on OpenRouter's docs page.
Yes — point the OpenAI SDK's base URL at `https://openrouter.ai/api/v1` and pass your OpenRouter API key.
Pay-as-you-go via prepaid credits. Most non-free models priced at OpenAI/Anthropic equivalents with a small markup; some cheaper than direct. If your traffic is bursty or seasonal, the free tier may be enough; if you need a guaranteed SLA, upgrade.