# Inference Index — Machine-Readable Description

type: data-service
category: ai-inference-pricing
domain: inferenceindex.ai

## Purpose

Real-time AI inference pricing comparison across major providers.
Useful for: cost optimization, provider selection, budget planning, automated procurement.

## API Endpoint

GET https://inferenceindex.ai/api/prices
Content-Type: application/json
Auth: none required
CORS: open (Access-Control-Allow-Origin: *)

### Response schema

{
  "updatedAt": "<ISO8601>",
  "cheapTokensDiscount": <float 0-1>,
  "discountPercent": "<string>",
  "models": [
    {
      "id": "<slug>",
      "name": "<string>",
      "tier": "Frontier|Performance|Budget",
      "details": {
        "contextWindow": "<string>",
        "privacy": "<string>",
        "notes": "<string>"
      },
      "providers": [
        {
          "name": "<string>",
          "input": <float>,
          "output": <float>,
          "url": "<string>",
          "bestPrice": <bool>,
          "live": <bool>,
          "discount": "<string>",
          "buyUrl": "<string>"
        }
      ],
      "cheapest": {
        "provider": "<string>",
        "input": <float>,
        "output": <float>,
        "savings": "<string>"
      }
    }
  ]
}

## Pricing units

All prices in USD per 1,000,000 tokens (1M tokens).

## Models tracked

Frontier: Claude Opus 4.6, Claude Sonnet 4.6, GPT-5.2, Gemini 3 Pro
Performance: Grok 4.1 Fast, Kimi K2 Thinking, DeepSeek V3.2
Budget: Llama 3.3 70B

## Providers tracked

Anthropic, OpenAI, Google, xAI, Moonshot, DeepSeek, Venice, CheapTokens

## Update frequency

Static prices: manual updates
CheapTokens prices: live (30s refresh)