- Organizations
- Peking University
Peking University: AI benchmarks & research
Peking University research groups behind Claw-Eval, PHYBench, Video SimpleQA, MLVU and STEM.
Maintained by Peking University
7 benchmarks- ClawEval-MM
ClawEval-MM is the multimodal variant of ClawEval, evaluating agentic problem solving with visual inputs.
- Leader
Qwen3.8-Flash-Next60.4%- Models scored
- 6
- MLVU
A comprehensive benchmark for multi-task long video understanding that evaluates multimodal large language models on videos ranging from 3 minutes to 2 hours across 9 distinct tasks including reasoning, captioning, recognition, and summarization.
- Leader
Qwen3.7-Plus87.4%- Models scored
- 10
- Leader
Qwen3 VL 32B Instruct82.1%- Models scored
- 9
- STEM
A comprehensive multimodal benchmark dataset with 448 skills and 1,073,146 questions spanning all STEM subjects (Science, Technology, Engineering, Mathematics), designed to test neural models' vision-language STEM skills based on K-12 curriculum. Unlike existing datasets that focus on expert-level ability, this dataset includes fundamental skills designed around educational standards.
- Leader
Qwen2.5-Coder 7B Instruct34.0%- Models scored
- 1
- VideoSimpleQA
VideoSimpleQA evaluates factual knowledge grounded in video content, measuring how accurately models answer short, fact-seeking questions about videos.
- Leader
Seed 2.1 Pro76.4%- Models scored
- 4