Skip to main content
ModelScale

Coding benchmark

DSBench-Hard leaderboard

DeepSeek DSBench Hard. Every model the catalog carries a published DSBench-Hard value for, ranked by that value.

CategoryCoding
MeasureProvider-reported score
TasksInternal hard coding-agent tasks
DifficultyAdvanced coding-agent challenges

DeepSeek's internal hard coding-agent benchmark.

DSBench-Hard ranking

2 models with a published DSBench-Hard value, ordered by that value, highest first. Models the source has not scored on this benchmark are not listed — they are unmeasured, not last.

Models ranked by their published DeepSeek DSBench Hard value
RankModelProviderProvider-reported score
1DeepSeek V4 Pro 0813DeepSeek67.2
2DeepSeek V4 Flash 0731DeepSeek59.6

Evidence key: Observed

Rows are ordered by the value DeepSeek V4 Flash 0731 update published, highest first. The source does not state whether a higher value is the better result, so this page does not either: for a benchmark that measures a rate of failure, read the table from the bottom.

All leaderboards · Coding capability leaderboard