Skip to content
Free$1 of credit for every new organization
All articles

Product

Introducing Usefully: one key for AI models and the live web

Usefully gives you models from Anthropic, OpenAI and Google plus web search behind one API key, with automatic failover, prepaid credits at provider prices, and per-request logs.

· 3 min read · Usefully team


Building with AI models today usually means juggling accounts: one key per provider, one bill per provider, a separate search API, and your own code to retry when a provider has a bad hour. Usefully replaces that with one API key, one prepaid balance and one set of logs, for models from Anthropic, OpenAI and Google and for live web search.

What Usefully is

  • An OpenAI-compatible API at https://api.usefully.info/v1: POST /v1/chat/completions and GET /v1/models. If your code uses the OpenAI SDK, you change the base URL and the key.
  • An Anthropic-compatible API at /v1/messages, so code written for the Anthropic SDK works too.
  • Models from three providers, named like anthropic/claude-sonnet-5-5, openai/gpt-5 and google/gemini-2.5-flash. Switching is a change of model ID.
  • Web search powered by Exa: results with title, URL, snippet and published date, optional full page text, and domain filters, on the same key.
from openai import OpenAI

client = OpenAI(base_url="https://api.usefully.info/v1", api_key=USEFULLY_API_KEY)
reply = client.chat.completions.create(
    model="anthropic/claude-sonnet-5-5",
    messages=[{"role": "user", "content": "Hello!"}],
)
print(reply.choices[0].message.content)

Reliability built in

Usefully tracks the health of each provider with health scores and circuit breakers. When a request fails because of the provider, it moves to the next one serving the same model, trying up to 3 providers within about 30 seconds. You're charged once, and failed requests aren't charged at all.

Pricing that's easy to explain

  • Tokens at the provider's list price, with 0% markup.
  • Prepaid credits, from about $5, in US dollars or Indian rupees. A 3.5% payment fee applies at top-up, not per request.
  • Each request reserves an estimate, is charged for actual usage, and releases the rest.
  • Your first organization gets $1 of free credit, with no card required.
  • Unused credit can be refunded under the refund policy. Payments are handled by Dodo Payments as merchant of record, including tax and invoices.

A dashboard for the whole team

  • Logs for every request: model, provider, tokens, cost and latency. Metadata only; prompt and response bodies aren't stored.
  • Spend by day, by key and by model.
  • Per-key limits: requests per minute, tokens per minute and a daily spend limit.
  • Low-balance alerts, organizations with members and roles, and a playground for trying models.

Security basics

API keys are shown once, stored as salted hashes, and can be revoked instantly. HTTPS is used everywhere. For sensitive data, the data_policy option routes a request only to providers with zero data retention or no training on your data.

Works with what you already use

Because Usefully speaks the OpenAI and Anthropic APIs, it works with their SDKs and with tools built on them, including LangChain, the Vercel AI SDK, LlamaIndex, Cursor and Continue, usually by changing a base URL.

Get started

  1. 1Create an account and an organization. The $1 credit is added automatically.
  2. 2Create an API key in the dashboard and copy it; it's shown once.
  3. 3Follow the quickstart, or try a model in the playground.

We built Usefully because we wanted it ourselves: fewer accounts, clearer costs, and less glue code between an app and the models it depends on. We'd love to hear what you build with it, and what's missing.

Browse the models, see pricing, or try web search. Questions: support@usefully.info. Partnerships: team@usefully.info.

More articles

Get started

Your first request is five minutes away.

  • $1 of free credit to start
  • No card, no subscription
  • Keep the SDK you already use
  • Real people at support@usefully.info
Introducing Usefully: one key for AI models and the live web · Usefully