AI Pulse.
What’s new in AI. What matters for your business.
Latest AI updates
Checks every 5 minutes while this page is open. Headlines and excerpts stay in the source language.
Which model fits your work?
Compare capabilities and API costs. A benchmark is a starting point; your own tasks are the real test.
Choose an order, then select a model to see its details. Models stay visible when benchmark scores are unavailable.
Loading current model data…
How to read this comparison
Prices are USD per million API tokens, not chat subscription prices. A token is a small piece of text. Context is the amount a model can handle in one request. Extra fees and provider pricing may vary.
Scores come from Artificial Analysis via OpenRouter. This table uses the Intelligence Index. Higher scores indicate stronger performance on that benchmark. Models without a verified score are marked Not rated. Saved scores show their review date and tested configuration. Tool support does not prove an agent is reliable, and popularity does not measure accuracy.
Start with the work.
Then choose the model.
A higher score won’t tell you whether AI fits your business. Start with one repetitive task and a clear definition of a good result. Compare a few models on the same examples, then check accuracy, cost and how much review is needed.
For agents, decide what they can access, what needs your approval and when they must stop. Those choices matter as much as the model.
Talk to Art about your workflow →