Provider Routing

AICredits routes every request to the right upstream provider based on the model ID you specify. This page explains the routing logic and how to control it.

Routing Logic

Explicit Routing

The most reliable way to route is to prefix your model ID with the provider name:

openai/gpt-4o            → OpenAI
anthropic/claude-sonnet-4.6 → Anthropic
google/gemini-2.0-flash  → Google
deepseek/deepseek-chat   → DeepSeek
mistral/mistral-large    → Mistral
xai/grok-beta            → xAI

Explicit routing guarantees the request goes to the provider you specify, with no ambiguity.

Heuristic Routing

If no provider prefix is given, AICredits uses heuristics to detect the provider from the model name:

Failover

When the primary provider is unavailable (circuit breaker triggered), AICredits automatically retries the request:

1. Healthy keys first — round-robin across healthy API keys for the same provider
2. Exponential backoff — 500ms, 1s, 2s between retries (max 3 attempts)
3. Fallback provider — if all direct provider keys are unhealthy, routes to an aggregated provider pool as last resort

The circuit breaker marks a provider key as unhealthy for 30 seconds after repeated failures, then recovers automatically.

Supported Providers

Model Discovery

To see all available models with their provider prefixes, use the models endpoint:

curl https://api.aicredits.in/api/models

Or browse Models → for a searchable list with pricing.