AI Model API Pricing Monitor — GPT, Claude, Gemini & 300+ models

Every AI model's API price, one dataset, always current — with automatic price-change detection.

AI providers change prices and ship new models weekly. Teams hand-maintain pricing spreadsheets that go stale in days. This actor replaces that: one run returns per-token pricing for 300+ models across OpenAI, Anthropic, Google, Meta, Mistral, DeepSeek, xAI and more — normalized to USD per 1M tokens so everything is directly comparable — and flags exactly which prices moved since your last run.

What it does

• 📊 Pulls from two structured sources: OpenRouter (live marketplace pricing) and LiteLLM's maintained database of official list prices
• 🔁 Normalizes everything to $ / 1M input tokens and $ / 1M output tokens
• 🚨 Price-change detection: each record is flagged none, new, or the exact delta (e.g. input $3.00 → $2.50/1M (-16.7%)) since your previous run
• 🔔 onlyChanges mode outputs just the moved prices — pipe it into Slack/email via Apify integrations for a free price-alert system
• 🤖 API, schedules, and MCP ready — AI agents can ask "what's the cheapest model with a 200k context window?" against live data

Who uses this

• Engineering teams choosing models on cost/performance — with real, current numbers
• FinOps & procurement tracking AI spend exposure as providers reprice
• Devtool & platform builders who need a pricing table that never goes stale
• AI agents doing autonomous model selection and cost optimization
• Newsletter writers & analysts covering the AI market — schedule daily, get every price move first

Output example

{
  "provider": "anthropic",
  "model": "anthropic/claude-sonnet-4.6",
  "inputPerMTokUsd": 3.0,
  "outputPerMTokUsd": 15.0,
  "contextWindow": 200000,
  "priceChange": "none",
  "source": "OpenRouter",
  "scrapedAt": "2026-07-14T15:04:05.000Z"
}

Input options

Field	Description	Default
sources	OpenRouter, LiteLLM, or both	both
providerFilter	e.g. ["openai","anthropic","google"]	all
modelFilter	e.g. ["gpt-5","claude","gemini"]	all
maxInputPricePerMTok	Price ceiling	none
detectChanges	Flag price moves since your last run	true
onlyChanges	Output only moved prices (alerting mode)	false

The killer workflow

1. Schedule this actor daily with detectChanges: true
2. Add an Apify integration (Slack, email, webhook) filtered on non-empty priceChangesDetected in the summary
3. You now know about every AI price cut and price hike the morning it happens — before your competitors reprice

FAQ

How is this different from checking provider pricing pages? One normalized dataset vs a dozen differently-formatted pages — plus history and change detection, which pricing pages don't give you.

Where does the data come from? OpenRouter's public models API (live routed marketplace pricing) and LiteLLM's continuously-maintained pricing database (official list prices). Both are structured sources, not fragile page scraping.

Why do some models appear twice? Marketplace pricing (OpenRouter) and official list price (LiteLLM) can differ — that spread is itself useful signal. Filter to one source if you want a single view.

---

Companion actor: GPU Cloud Price & Availability Tracker — same normalized approach for GPU rental prices.