Benchmark

DSBench-Hard

DSBench-Hard is DeepSeek's internal test set of difficult coding-agent problems.

Modality
text
Categories
agents, code
Openness
unknown
Source
llm_stats
Reported scores
2

Reported scores

llm_stats

ModelOrganizationReported valueReportedEvidence
DeepSeek-V4-Flash-0731DeepSeek0.5962026-07-31evidence
DeepSeek-V4-Pro-0813DeepSeek0.6722026-08-13evidence

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.