Every AI model.
One endpoint.
ScoutCode routes your requests across 10 AI providers behind a single OpenAI-compatible API. When one model runs out of quota, the next one answers.
- 429groq / llama-4-maverickrate limited, cooling 38s
- stallgoogle / gemini-3-flashno first token in 8s
- 200mistral / mistral-medium-3streaming · 412 tok/s
Up and running in a minute
- 1
Create an account
Sign up and make an API key on your account page. Keys can be named, paused and revoked any time.
- 2
Change the base URL
Point any OpenAI client at http://localhost:4321/v1. Your existing code, SDKs and tools keep working.
- 3
Pick a model, or let us
Ask for a specific model, or send "auto" and ScoutCode routes each request to the best one available.
Built for apps that can't go down
Rate limits, outages and slow models are handled for you, so your users never see them.
Automatic fallback
When a model hits a rate limit, errors or stalls, the next one takes over, even mid-stream before the first token.
Speed Router
Rates how hard each request is, then picks the model and reasoning effort to match. Quick questions stay quick.
Streaming
Server-sent events exactly like OpenAI, so chat UIs and coding agents render tokens as they arrive.
Tools and vision
Tool calls and images are routed only to models that support them, and translated between providers.
Key rotation
Requests spread across many upstream keys. Rate-limited keys cool down for exactly as long as the provider asks.
Clear limits
Per-account request limits, visible on your account page, with OpenAI-style 429s when you reach them.
Works with the OpenAI SDK you already use
Change two lines: the base URL and the key. Streaming, tool calls and images work as they do on OpenAI.
curl http://localhost:4321/v1/chat/completions \
-H "Authorization: Bearer sk-scout-…" \
-H "Content-Type: application/json" \
-d '{"model":"auto","messages":[{"role":"user","content":"Hello!"}]}'from openai import OpenAI
client = OpenAI(base_url="http://localhost:4321/v1", api_key="sk-scout-…")
res = client.chat.completions.create(
model="auto",
messages=[{"role": "user", "content": "Hello!"}],
)
print(res.choices[0].message.content)import OpenAI from 'openai';
const client = new OpenAI({ baseURL: 'http://localhost:4321/v1', apiKey: 'sk-scout-…' });
const res = await client.chat.completions.create({
model: 'auto',
messages: [{ role: 'user', content: 'Hello!' }],
});
console.log(res.choices[0].message.content);Popular models
Questions
What is ScoutCode?
An AI model router. It sits in front of many providers and exposes them as one OpenAI-compatible API, handling fallback, key rotation and capability matching for you.
Do I need to change my code?
No. Set the base URL and API key in any OpenAI client and keep using the Chat Completions API you already know.
What does "auto" do?
It tries a ranked list of models and moves to the next one whenever a model is rate limited, fails or stalls, so a single provider outage does not break your app.
Which models can I use?
Everything on the Models page, from free-tier models on Groq, Gemini, Cloudflare and others to paid frontier models. Each one lists its context length, capabilities and price.
How do limits work?
Limits belong to your account and are shared by all its keys. During early access, new accounts are activated by hand, and your account page shows your current limits.
Stop babysitting rate limits
One key, every model, automatic fallback.