Contact
Get your model on the leaderboard, sponsor a run, suggest a model, or ask about the results. Email is the fastest way to reach us.
Email[email protected]Get your model on BB Arena
- AI labs and model providersEvaluate your modelHave a public or upcoming model run through all 100 labs under the standard protocol, and published alongside every other model.Request an evaluation →
- Open-weight teams, hosting providers, anyoneSponsor a runCover the API costs for a model you want on the leaderboard. Useful for open-weight models and new providers.Ask about sponsoring →
- EveryoneRequest a modelTell us which model or effort level you'd like to see next. Requests help us decide what to run.Suggest a model →
How an evaluation works
- Get in touch with the model, the effort levels you'd like tested and how to access it.
- Access check. We confirm the model works in our harness with a short test on a few labs.
- Full run. All 100 labs, both attempts, under exactly the same protocol as every other model. A run usually finishes within a few hours.
- Results published with the same detail as every model: solves, rungs, cost, speed and results by vulnerability class.
What we'll need
- API access to the model, with tool (function) calling
- The model name and the effort levels to test
- Its context window, and published per-token prices so we can report cost
- Rate limits that allow several agents to run at once
Same rules for everyone. Every model runs in the same harness, under the same protocol and scoring, whoever asked for it or paid for it. Sponsoring a run covers its running costs; it doesn't change how the model is tested or scored. See the methodology for exactly how.
Everything else
- CorrectionsQuestion a resultThink a figure is wrong or a price is out of date? Tell us which model and why.Report an issue →
- Media and academiaPress and researchInterviews, analysis, or using the results in your own work.Get in touch →Citing BB Arena
- Responsible disclosureSecurityFound a security issue with this site? Please report it privately by email.Report privately →