72 models · 16 vendors

One API for every model you actually use.

Claude, GPT, Gemini, Grok, DeepSeek and more behind a single OpenAI-compatible endpoint. Pay per request from prepaid credit — no subscription, no minimum, no renegotiating when you switch model.

Create an account Browse models and prices

No card required to look around. Prices below are the real ones.

Estimate a month

Live prices

ModelRequests per month10,000Input tokens per request1,000Output tokens per request500

Your monthly spend

$1.58

No subscription. You are charged per request, from prepaid credit.

Change two lines, keep your code

The API speaks OpenAI's dialect — streaming, tool calls and usage reporting included. Point your existing client at us and the rest of your application stays exactly as it is.

· Server-sent events, relayed as they arrive
· Token usage on every response
· Errors in the envelope your client already handles

Cursor, Cline, Aider and more Ecosystem

from openai import OpenAI

client = OpenAI(
    base_url="https://api.aiomodels.dev/v1",
    api_key="sk_live_...",
)

completion = client.chat.completions.create(
    model="claude-opus-5",
    messages=[{"role": "user", "content": "Hello"}],
    stream=True,
)

Priced per million tokens

A sample of the catalog. These are live prices, not examples.

All 72 models

Model	Input	Output
GPT 5.6 Luna`gpt-5-6-luna`	$0.012	$0.072
DeepSeek V3.2`deepseek-v3-2`	$0.129	$0.195
Qwen3.5 Flash`qwen3-5-flash`	$0.20	$0.20
MiMo V2.5`mimo-v2-5`	$0.101	$0.202
Claude Haiku 4.5`claude-haiku-4-5`	$0.045	$0.225
GPT 5.6 Terra`gpt-5-6-terra`	$0.228	$0.228

Built like infrastructure

One API, every model

OpenAI-compatible, so an existing integration switches over by changing two settings. No client library to replace.

Prepaid, not subscribed

Top up and spend it. No monthly minimum, no seat count, no commitment to renew something you stopped using.

Automatic failover

Every model is served by a ranked chain of providers. If one degrades, the next takes the request — without a code change on your side.

Priced per request

Every call is recorded with its tokens, latency and exact charge. Your balance is a ledger you can audit, not an invoice you receive later.

Keys stay yours

We hold a hash of your API key and nothing more. Provider credentials are ours to manage — you never handle them.

Stable model names

You address models by our identifiers. We can change supplier behind one without touching your code or your price.

Start with a few dollars

Create an account, top up, and make your first request. Whatever you do not spend stays on your balance.

Create an account See pricing