Benchmark Almanac

Benchmarks Arenas

Leaders Models Compare Labs Index

Model cards Reporting

Compare Model cards Reporting New this month Audits Sources Contribute

/

Labs

The University of Hong Kong (XLANG Lab)

2 models 0 model cards read xlang.ai

Every The University of Hong Kong (XLANG Lab) model

OpenCUA-32B Jul 2025
OpenCUA-7B Jul 2025

Benchmarks released each month since 2019, colored by area.

Benchmark Almanac

Scores from lab launch posts, system cards and 1,747 public leaderboards.

3,324 benchmarks, 1,170 models and 104 labs. Data as of November 14th, 2026.

Built and maintained by Brian Zhou.

Submit a benchmark

Explore

Benchmarks
Arenas
Leaders
Models
Labs
Compare models
Model cards
What labs report
Build your own index
New this month
Audits
Contribute

Areas

Knowledge and language
Math and science
Coding
Agents and tools
Professional work
Vision, audio and robotics
Safety and security

Method

How we compare scores
Lab claims next to leaderboards
What gets cataloged
Where the scores come from
In use, fading, dropped
Versions
Who held number one
Finding new benchmarks
Known issues
Sources

Scores from Epoch AI and Arena are under CC BY 4.0, and scores from other leaderboards are under each board’s own terms.