Deepseek V4 — 15.9% on BenchFae

Deepseek V415.9% overall on BenchFae, ranked 12 of 15.

Category

Score

Automated

Fairy

Human

Full-Stack Web Apps

22.7%

22.8%

26.2%

19.2%

2D Game Development

17.8%

24%

19.4%

14.3%

3D Game Development

6.4%

18.8%

14.3%

0%

Website Frontend

11%

19.5%

16.5%

0%

Website Backend

16.2%

15.5%

17.1%

—

Debugging/Refactoring

22%

25.1%

18.3%

—

Adversarial Code Review

15.2%

15%

15.4%

—