Skip to main content
The FlexAI Inference API is an OpenAI-compatible HTTP gateway for text, code, reasoning, vision, and embedding models. Point any OpenAI SDK at https://api.flex.ai/v1 and your existing code works unchanged.

Quickstart

Get a key and make your first request in under two minutes.

Model catalog

Browse every model we host, with context windows and pricing.

Model discovery

Filter the live catalog from code and find the right endpoint per modality.

Streaming

Stream tokens token-by-token with usage tracking.

Vision

Send images to multimodal models in the OpenAI image_url shape.

Tool use

Call functions from model responses.

​
What you get

  • One API key, every model. Chat, completions, vision, and embeddings all authenticate with the same sk-… bearer token.
  • Account-level credit, per-key rate limits. Your balance is shared across every key on the account. The rate limit is set on the account, but each key is counted against it separately rather than sharing one window. Exceed either and you get an unambiguous 402 or 429 with headers explaining why.
  • No card to start. Sign up and mint a key without entering payment details. Add a card when you’re ready to spend; usage is then metered at per-model rates, with no token packs and no subscriptions. Prices and usage are in dollars everywhere; EU accounts are charged and invoiced in euros (see Regions).
  • Pay as you go. No commitment beyond your usage, so you can start evaluating at whatever scale suits you and stop whenever you like.

​
OpenAI compatibility

The surface is intentionally identical to OpenAI’s for the endpoints we support. If your code runs against api.openai.com, swap the base_url and the model id — that’s the whole migration. See the compatibility matrix for what’s supported and what isn’t.

​
Where things live

Which API host you use follows the region your workspace was created in, and a key is only valid against its own region’s host. If you haven’t chosen a region, you’re in us and https://api.flex.ai/v1 is yours. See Regions. Use https://api.flex.ai/v1 for anything new in the US region, and https://eu.api.flex.ai/v1 if your workspace is in the EU. Those are two different stacks serving different workspaces, not two names for the same one — a key is valid against one of them, not both. The dashboard is a browser app, reached at platform.flex.ai and never on an API host, so the two kinds of link are not interchangeable. The host is the only thing that changes in how you call us. Paths, request bodies, headers and response shapes are identical on either:
Swap /v1/models for /v1/pricing to read the price list, or /v1/chat/completions to make a request — see the quickstart. What the two regions do not share is which models they serve: ask the host you are calling rather than porting a model id across, as what each region serves explains. It is worth setting the host once in config rather than per call. If you integrated with us early and your code calls a FlexAI host that is not listed above, it keeps working. Nothing here deprecates it and there is no migration to do — these are the hosts to use for new work, not a replacement you need to move to. If you are unsure which one your key belongs to, ask us at support@flex.ai rather than guessing: a key sent to the wrong host returns 401, which looks like a bad key.