Benchmark

PaperBench

PaperBench is a benchmark for evaluating AI agents on their ability to replicate research papers. It tests models on complex, multi-step workflows…

Modality
text
Categories
reasoning, agents, code
Openness
unknown
Source
llm_stats
Reported scores
3

Reported scores

llm_stats

ModelOrganizationReported valueReportedEvidence
Kimi K2.5Moonshot AI0.6352026-01-27evidence
MiniMax M3MiniMax0.5262026-06-01evidence
Qwen3.8 MaxQwen0.932026-08-02evidence

Scores are partitioned by the source that reported them and are never merged into a single cross-source ranking, because the sources measure different things and say so.