Hardware leaderboard
Each computer's best run on any model. Higher is better: fast and correct. How the score works.
Compare two machines on the models both ran.
Your computer
Choose your computer to see where it stands among the same machines.
Everyone
| # | Score | Machine | Model | tok/s | Tasks | Tokens | By |
|---|---|---|---|---|---|---|---|
| 1 | 2,256Silver | M4 Max 128 GB | Qwen3.8-27B-Fast1 run · llama.cpp · resyn.cloud in the browser | 23.4 | 49/50 | 100k | uwu |
Scrolls sideways to show every column.
Tiers: Bronze 0+ · Silver 1,500+ · Gold 3,000+ · Platinum 6,000+ · Diamond 10,000+. Ties go to more points, then more tasks passed, then the earlier run. A score ranks when a signed-in account posts it on a named computer (picked from the list or typed in), answers every scored task through a model's API, and its speed looks measured. A run ranks at once, with no second run needed. A run faster than its computer can read the model, or far faster than the others on it, is held for review. Up to ten runs a UTC day per account rank, across its computers and models. No more than five runs of one model on one computer a UTC day rank. The rest are shown with the reason. Each entry says how many runs its account posted of that model on that computer.