Which AI should you use right now?
Pick what you're doing. We rank the frontier on objective telemetry and community input.
⌕
What just shipped
The latest from each lab
One score, many sources
one score · lightly community-adjusted
The infotracker Index
◆ How the Index is scored — weights, sources, methodology
Deep dives
Explore
Cost per good result, not per token — what you actually pay to finish the job. Pick the kind of work; everything below reprices for it.
Every model, priced honestly per 1M tokens · plus cost per good result
| Model | Blended $/1M | 1st-try | Tries | $ / good result | Value index |
|---|
Real-world throughput — median output speed and first-token latency.
Is it up right now? Provider health, uptime and recent incidents.
Shipped — releases, outages & pricing changes.
Unconfirmed — leaks and speculation, each with a rough chance of being true.
Community Elo — per-category ratings, nudged by your top-5.
♟ How is Elo calculated? — what goes into this number
The full roster — every model we track, proprietary and open.