It scores 64.6 on the leading intelligence test. Second place scores 62.1
Try it- 2GPT-6.2 Astra2.5 points behind
- 3Gemini 4 Neon3.2 points behind
We check 376 AI models every day and tell you in plain words which one is the smartest, which one gives you the most for your money, and which one people use the most
It scores 64.6 on the leading intelligence test. Second place scores 62.1
Try itAbout 8 times cheaper than the smartest one, and still 79% as smart
Try itApps and developers used it 1.3 times more than the runner-up this week
Try itEvery day you open the board while logged in fills one square
New models, price moves and new number ones, day by day
Different models win at different work. This is today's number one for each job
Pick the job and what you are willing to pay. We show the best three that fit
Nothing in this job's top group fits that budget. Try a higher one
Job rankings come from Design Arena, where people vote on results without knowing which AI made them.
Free account. We only write when something changes: a new number one, a price drop on a model you follow, and a short weekly summary
Full rankings, prices and a chart, for anyone who wants the detail
Updated Nov 19, 2026 — 376 AI models tracked.
The newest models from the last 30 days
Every dot is a model. Higher is smarter, further left is cheaper. The green ones are the only models nobody beats on both at once
Search, filter and sort everything. Prices are US dollars per 1 million tokens
| My watchlist |
|---|
Intelligence is the Artificial Analysis Intelligence Index, higher is smarter. Value is intelligence points per dollar. Weekly use is tokens (pieces of words) processed through OpenRouter. Speed is tokens written per second. Memory is how much text the model can read at once, in tokens.
Everything on this page is free to read. Pro is for people who pick AI tools for their work
$0
$12 a yearThat is 3 cents a day
Pro opens soon
Prices, model details and usage come from OpenRouter, a service that sends requests to hundreds of AI models and publishes how much each one is used. Intelligence scores are the Artificial Analysis Intelligence Index, an independent benchmark that turns many hard tests into one number.
Once a day, automatically. The time of the last update is shown on the board at the top.
The amount of text each model processed through OpenRouter. It is real traffic from real apps and developers. It only covers OpenRouter, so people chatting inside ChatGPT, Gemini or Claude's own apps are not counted.
The model that gives the most intelligence for every dollar. To keep weak models from winning only by being cheap, a model has to be at least 60% as smart as the current leader to qualify.
Yes. Everything on this page is free to read. A free account adds a weekly email and a watchlist. Pro adds same-day alerts and the full ranking for every job.
No. Rankings are calculated from data only. Anything sponsored is clearly labeled and sits outside the rankings.