Introduction
What AivoKey is and why it exists.
AivoKey is a single API in front of many AI providers. You get one endpoint, one key, and one bill while the router picks the cheapest healthy provider for every request. It speaks three wires so any client works unchanged:
- Chat Completions —
POST /v1/chat/completions([OI]-compatible) - Responses —
POST /v1/responses([OI]-compatible) - Messages —
POST /v1/messages(Anthropic-compatible)
Why AivoKey
- Bare model names. Call models by their real name —
deepseek-v4-pro,gpt-5.5— with no provider prefix. One name maps to every provider serving that model. - Automatic routing. Requests go to the cheapest healthy provider, with cooldown and failover across distinct providers when one is unhealthy. Provider names are never exposed to your client.
- Budgets that fit. Plans grant specific models with 5-hour, weekly, and monthly windows; PAYG draws from a prepaid balance. A plan only serves the models its operator allows.
- Prompt-cache aware. Session affinity keeps a conversation on one provider so provider-side prompt caching keeps working.
Base URL
Every request goes to:
https://api.aivokey.dev
The [OI]-compatible wires use https://api.aivokey.dev/v1/...; the Anthropic client appends /v1 itself, so configure its base URL as https://api.aivokey.dev (without /v1).
Next steps
- Quickstart — send your first request.
- Authentication — how keys work.
- Models — the live catalog.
- Plans & billing — what each plan includes.