Docs
Docs
Models
Chat
Compare
Providers
Apps
Rankings
Search
Ctrl K
Sign Up
Sign Up
DSBench-Hard Benchmark Leaderboard | Phaseo
DSBench-Hard
Overview
Overview
Type: percentage
Code
Recorded Results
1
Average Score
59.60%
Score Range
59.60% - 59.60%
Leading Model (lowest score)
59.60% - DeepSeek V4 Flash 0731
Scores Over Time
Individual benchmark scores plotted by date.
Models Using This Benchmark
Organisation
Model
Reported
Top Score
Info
Self Reported
Source
DeepSeek
DeepSeek V4 Flash 0731
31 Jul 2026
59.60%
Internal difficult coding-agent test set; DeepSeek-V4-Flash-0731
Yes
Source
