Model & skill benchmarks
Every evaluation is run by SinoBee and published in full: time, proposer, skill, model, method, process and result. Anyone can crowdfund an evaluation; the report is published once the backers accept it.
Model leaderboard
| # | Model | Overall | Benchmark | Rubric | A/B | reports |
|---|---|---|---|---|---|---|
| No evaluation reports yet. Be the first to propose one. | ||||||
Scores are the average of reports without a skill (0–100). 12 models are offered.
Latest reports
No evaluation reports yet. Be the first to propose one.
Propose
Describe the model or skill to evaluate; AI writes the method, criteria and cost estimate, and people back it.
Run
Once funded, SinoBee runs the evaluation in an isolated container, recording every step, and the backers review it.
Publish
The report is public here and on the skill page for anyone to read and discuss.