Models & Providers

AICredits supports 300+ models across multiple providers through a single API.

Supported Providers

Model ID Format

Use the provider/model-name format for unambiguous routing (recommended):

"model": "openai/gpt-4o-mini"
"model": "anthropic/claude-sonnet-4.6"
"model": "google/gemini-2.5-flash"
"model": "deepseek/deepseek-chat"

You can also pass just the model name (for example gpt-4o-mini or claude-haiku-4.5). Well-known prefixes such as gpt, claude, gemini, deepseek, mistral and grok are matched to their provider, and unrecognised names are served from the aggregated provider pool.

Provider Routing

Each request resolves to a primary provider and an ordered list of secondary routes. Where a model is available from several providers, the routes configured for it are tried in priority order, with an aggregated provider pool as the last resort. The primary provider uses round-robin key selection across healthy keys, and the fallback is automatic and transparent.

You can steer this per request with provider.order and provider.allow_fallbacks. See Provider Routing for details, and check the X-AICredits-Provider response header to see which provider served a call.

Circuit Breaker

AICredits tracks the health of each provider key. If a key fails multiple times, it's temporarily marked as unhealthy and skipped for 30 seconds. This prevents repeated failures from adding latency to your requests.

Model Fallback

Send a models array instead of a single model to enable automatic cross-model fallback. If the first model fails (5xx, rate limit, or model unavailable), the request is automatically retried on the next model — transparently, with no extra code in your application.

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.aicredits.in/v1",
  apiKey: "sk-your-aicredits-key",
});

const response = await client.chat.completions.create({
  // @ts-expect-error — AICredits extension
  models: ["openai/gpt-4o-mini", "anthropic/claude-haiku-4.5", "google/gemini-2.5-flash"],
  messages: [{ role: "user", content: "Hello!" }],
});

// Check which model actually responded
console.log(response.model);

from openai import OpenAI

client = OpenAI(
    base_url="https://api.aicredits.in/v1",
    api_key="sk-your-aicredits-key",
)

response = client.chat.completions.create(
    # models[] is an AICredits extension — pass as extra_body
    model="openai/gpt-4o-mini",  # required by SDK; overridden by models[] below
    extra_body={
        "models": ["openai/gpt-4o-mini", "anthropic/claude-haiku-4.5", "google/gemini-2.5-flash"]
    },
    messages=[{"role": "user", "content": "Hello!"}],
)

print(response.model)  # which model actually responded

curl https://api.aicredits.in/v1/chat/completions \
  -H "Authorization: Bearer sk-your-aicredits-key" \
  -H "Content-Type: application/json" \
  -d '{
    "models": ["openai/gpt-4o-mini", "anthropic/claude-haiku-4.5", "google/gemini-2.5-flash"],
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

Browse Models

View all available models with real-time pricing on the Models page, or query the API:

# Public model catalog endpoint (outside OpenAI /v1 base URL)
curl https://api.aicredits.in/api/models

Deprecated model redirects

When a provider retires a model, AICredits may redirect requests for the old ID to its replacement so your app keeps working. A redirected request is billed at the replacement model's price, and the response tells you what happened:

The model field in the response body is the model that served the request. If you need a specific model and want a request to fail instead of being redirected, check X-Redirected-From in your client and alert on it.