Leaderboard.cn is an AI model ranking and evaluation reference for the Chinese-speaking community. It currently focuses on large language, open-weight, and multimodal models, organizing results from provider reports, public leaderboards, and model cards so readers can quickly understand relative performance across benchmarks.
Traceable data
We retain sources, snapshots, report names, and original scoring conventions wherever possible, so results can be checked against their origins.
Cross-source view
Results scattered across reports and leaderboards are organized in one place and can be explored by model, benchmark, or organization.
Careful comparisons
Benchmarks with the same name can use different protocols across reports. Results shown here are not absolute conclusions about capability.
What we cover
- Core model information, including release date, organization, weight availability, parameter count, context length, and official links.
- Benchmark results from public reports and leaderboards, including sources, snapshots, and original notes.
- Pricing from common API providers, preserving the original billing currency and unit wherever possible.
- Domain-specific frontier rankings calculated from the evaluation results in our dataset.
What this site is not
Leaderboard.cn is not a general product-ranking site and does not claim to cover every AI product, application, or tool. There is currently no committed timeline for expanding into a general product ranking. Rankings reflect only relative performance within the data we have collected; they do not represent final performance in every real-world setting.
About the rankings
Frontier rankings are calculated from relative placements in collected benchmarks and remain subject to data selection, protocol differences, and coarse reliability weighting. For the full method and its limitations, see Ranking methodology.