Frontier models, far below list price.

GPT, Claude, Gemini and Grok for a fraction of what the labs charge. Change the base URL and keep your code.

Prepaid from $20 · No subscription · OpenAI- and Anthropic-compatible

Today’s pricesLoading prices…

List price vs. Infer.

Infer’s lowest open rate for each model next to the lab’s published API price, per 1M tokens. Pulled from the live catalog when this page loads.

Loading current prices…

List prices are the labs’ published API rates, checked on the date shown in each model’s details. The saving is the smallest one across input, cached input and output on the route shown. Compare every model

Choose your discount, request by request.

Every model ID ends in a route. Cheaper routes draw on a wider pool of qualified supply; stricter routes accept fewer suppliers and cost more. Prices below are for the model selected above.

:low-cost

Low cost

The deepest discount. Each request goes to the lowest-cost supply that passes Infer’s checks for the model.

Input / 1M—
:balanced

Balanced

Supply has to pass stricter health, capability and privacy checks first. The lowest-cost supply that clears them serves the request.

Input / 1M—
:stable

Stable

The strictest checks short of Official. Fewer suppliers qualify, so you pay more for consistency.

Input / 1M—
:official

Official API

The lab’s own API, billed through your Infer wallet at 1.25× the lab’s list price.

Input / 1M—

Routes change suppliers, never the model family. Supplier identities stay private. How routing works · Current status

Change the base URL. Keep your code.

Infer speaks the OpenAI Responses and Chat Completions APIs and the Anthropic Messages API. Streaming and tool calls work wherever the model supports them. The samples use the model and route selected above.

curl https://infer.flow7.org/v1/responses \
  -H "Authorization: Bearer $INFER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "infer/gpt-5.6-terra:low-cost",
    "input": "Summarize this support ticket in two sentences."
  }'

Prepaid, capped and itemized.

You pay the published rate for what you use. There’s no subscription and no top-up fee.

Prepaid wallet

Add $20 or more and spend it on any model and route. When credit runs out, requests stop, so there’s no invoice to surprise you.

Limits on every key

Give each app or teammate its own key with daily and monthly caps. Requests already running count toward the cap.

A record for every call

Each completed request lists its model, route, tokens and charge, in your dashboard and in the API response.

The short answers.

Anything else, ask support or read the docs.

Why are prices below the lab’s list price?

Infer sources each model from authorized suppliers and sends every request to the lowest-cost one that passes the route’s checks. Official API is the exception: it uses the lab’s own endpoint at 1.25× list.

Does my request run on the model I chose?

Yes. A route can change the supplier, never the model family in your model ID. Supplier identities stay private.

Which route should I start with?

Low cost, for the deepest discount. Move a workload to Balanced or Stable when it needs stricter checks, or to Official API when it needs the lab’s own endpoint. Switching is a one-word change in the model ID.

What happens if no supplier can take my request?

Infer returns an error before the request runs, and you aren’t charged. Your route and its price never change on their own.

How does billing work?

Prepaid. Add $20 or more to your wallet. Each request is charged at the published rate for its model and route, and unused credit stays in your wallet.

What can I connect?

Anything that uses the OpenAI Responses or Chat Completions API, or the Anthropic Messages API: the official SDKs, LangChain, the Vercel AI SDK, Pydantic AI, and tools such as Codex, Claude Code and OpenCode.

Start paying less per token.

Sign up free, compare every model and route, and add credit when you’re ready.

Sign up → Create key → Add funds → Send a request