Miavo
Public beta

Every AI model, one Miavo key.

Browse models, test prompts, create a key, add credits or bind a wallet, then ship through the OpenAI SDK.

No credit cardDrop-in for OpenAI SDKStreaming + tools
quickstart.ts
import OpenAI from 'openai';

const client = new OpenAI({
  baseURL: 'https://api.miavo.xyz/v1',
  apiKey: 'sk-maas-...',
}); 

const response = await client.chat.completions.create({
  model: 'claude-sonnet-4-6',
  messages: [{ role: 'user', content: 'Write me a haiku.' }],
});
How it works

Built like a backbone, not a wrapper.

Smart routing

One key, many providers. Miavo selects a healthy upstream from a managed pool with per-credential locking, automatic failover, and live quota tracking.

Encrypted at rest

Every credential — OAuth token, API key, cookie — goes through ChaCha20-Poly1305 with a master key only your server knows.

OpenAI-compatible

Drop Miavo in front of any OpenAI SDK. Streaming, function calling, multi-modal — translated transparently to each backend.

Usage you can read

Tokens, latency, cost — every request logged per key. Set RPM and TPM limits per key, not per account.

OAuth that refreshes itself

Background workers refresh Claude and Gemini OAuth tokens before they expire. Pool keeps serving while you sleep.

Pre-paid credits

Top up once. Spend transparently per-token across every provider. No vendor lock-in, no surprise bills, no need to manage upstream accounts.

Models, on day one.

More providers land each month. The adapter pattern is open — bring yours.

See full catalog
Anthropic
Claude Opus 4.7
Anthropic
Claude Sonnet 4.6
Bedrock
Claude via Bedrock
Gemini
Gemini 3.1 Pro
Gemini
Gemini 3.1 Flash-Lite

Ship in five minutes.

Sign up, copy a key, change your baseURL. That's it.