We rank models by intelligence per dollar — priced at the cheapest major provider actually selling them, not at the lab’s list price.
We see prices for 1,768 models and list the 347 that pass our filters — no deprecated base models, dated snapshots or preview builds padding the count. See exactly what we drop and why →
Browse the full model directory with live pricing, check today’s cheapest LLM API, compare intelligence per dollar, compare any two models head-to-head, jump into any of our 117 cost calculators, compare the developer AI stack — vector databases, agent frameworks, observability and more — or see AI tool pricing for Cursor, Copilot, ChatGPT and Claude.
Explore 347 AI models
Prices and Arena intelligence for every model we track — filter by provider, price, context and capability.
Who got cheaper, by how much, and when — something a chatbot with stale training data can’t tell you.
- -60%
Qwen3.8 2.4T A95BCached input $0.500 → $0.200/1M
- -1%
Qwen3.8 2.4T A95BOutput $6.00 → $5.95/1M
- -3%
Qwen3.8 2.4T A95BInput $2.00 → $1.95/1M
- +400%
Qwen3.8 27BCached input $0.010 → $0.050/1M
- -13%
Qwen3.8 27BOutput $0.400 → $0.350/1M
- -20%
Qwen3.8 27BInput $0.100 → $0.080/1M
Jump straight to a provider’s models and prices — no digging required.
Popular calculators
Browse by category
Frequently asked questions
Which LLM API is the cheapest?+
It depends on your input/output mix, but small models are usually cheapest — right now the lowest blended rate we track is about $0.010 per 1M tokens. See the LLM prices today table for the full cheapest-first ranking.
How much do GPT, Claude and Gemini cost per 1M tokens?+
Each provider prices input and output tokens separately, and output is usually 3-5x more expensive. Open each model's page for its exact input, output and cached rates, or use the LLM API cost calculator to turn those rates into a per-request and monthly bill.
How often are the prices updated?+
Prices are re-checked daily against each provider's official pricing and the site is rebuilt automatically, so the numbers you see are current for today. This page's data was last verified 18 September 2026.
How do you measure how smart a model is?+
We show LMArena (arena.ai) Elo scores, a crowd-sourced leaderboard from millions of blind human head-to-head votes. It is the single, clearly attributed intelligence source we use; models with no Arena score simply show no score rather than a guess.
Are the open-weight model prices first-party?+
Where a lab publishes an official API price (OpenAI, Anthropic, Google, Mistral, and most Qwen/Kimi/GLM models) we use it. For open-weight models a lab doesn't sell directly, we show a clearly labeled representative host rate so the model can still be compared.
Do I need to sign up or install anything?+
No. Every calculator and table runs in your browser with no signup, and results update live as you type. You can copy a shareable link to any calculation or filtered view.
Why AI Calculator Pro?
Instant & simple
Sensible defaults and live results. Get an answer in seconds.
Transparent
Every calculator shows the math and links to official pricing.
Cross-provider
Compare OpenAI, Claude, Gemini, DeepSeek and more in one place.