Direct to the lab · not an aggregator

Best models.
Lowest latency.
One key.

Aggregators like OpenRouter bundle every model behind one key — then route you through extra hops. Float is a single-hop gateway to DeepSeek, Qwen, Kimi, GLM, MiniMax, and Mistral: one API key, one prepaid balance, pass-through lab pricing. Swap model when something better ships; your app stays the same.

  • Lowest latency — single hop to the provider, not 100ms+ aggregator routing
  • Best per dollar — direct lab pricing; follow the frontier as models change
  • Indie-simple — one key, one balance, OpenAI-compatible
  • From anywhere — same API worldwide; gateway colocated with labs

How it works

Same convenience as an aggregator — without the extra hop. One key, any model, minimum latency.

Today

One key, any model

You
Float
DeepSeek, Qwen, Kimi…

One API key · every coding model

Use one API key for all your coding models. Change model to switch providers — no new keys, no new billing accounts.

Next

Your key works everywhere

You
Provider API

No middleman on the request

Provider
Float

Billing happens after — not during — the request

Providers accept the same Float key directly. You talk to them at full speed; we handle the bill in the background.

Target providers

Six labs on the roadmap — all on the same API key. Use provider/model in requests once an integration is live.

Loading providers…

For developers

Built for indie devs who want the best model today — fast, cheap, and dead simple. OpenAI-compatible; change model, not your integration.

TypeScript
import OpenAI from "openai";

const client = new OpenAI({
  apiKey: "ccai_...",           // one API key for all coding models
  baseURL: "https://float-api.com/v1",
});

const res = await client.chat.completions.create({
  model: "deepseek/deepseek-chat",  // or qwen, kimi, glm…
  messages: [{ role: "user", content: "Ship it." }],
});

Early access — contact partners@api-float.com for an API key.

Are you a model lab?

We're seeking reseller and native-key partnerships. Share your integration spec in one click.

View partner page →