Skip to content

Usefully ​

Usefully is one API for two things most AI products need: LLMs from many providers and web search, behind one key and one prepaid balance.

  • OpenAI-compatible. Point the OpenAI SDK (or any OpenAI-compatible tool) at https://api.usefully.info/v1 and call any model, including Claude and Gemini. Anthropic's Messages API works too, at /v1/messages.
  • Automatic fallback. When a provider errors, overloads or times out, the request moves to the next one serving that model, before you see a failure.
  • Pay per request, at provider prices. Token prices are passed through unchanged. Credits are prepaid; there are no subscriptions.
  • One bill, one key, one log. Every request's cost, latency and serving provider is in your dashboard.
bash
curl https://api.usefully.info/v1/chat/completions \
  -H "Authorization: Bearer $USEFULLY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "anthropic/claude-sonnet-5-5", "messages": [{"role": "user", "content": "Hello!"}]}'

Next: make your first request in five minutes.

What's available ​

CapabilityEndpointModels
Chat / LLMsPOST /v1/chat/completions, POST /v1/messagesOpenAI, Anthropic, Google, Llama (via Groq and Together)
Web searchPOST /v1/searchusefully/web-search
Reading pagesPOST /v1/extractusefully/extract
CatalogGET /v1/modelsLive list with prices; no key needed

More capabilities (embeddings, crawling, deep research and others) are planned for later releases.

See the full, current list at GET /v1/models.