VERIFYAI infra
Jev AI Judging Performance
Alex Varga /✦@vargastartup
0:41Jev is an AI judging system that demonstrates significantly faster and cheaper performance compared to Qwen3 Max, while maintaining high agreement on verdicts.
JEV is INSANE at judging:
- Qwen3 Max (reasoning off): 2h 11m
- @typesafeai Jev: 4m 20s
30x faster, 94x cheaper ($3.39 vs $0.036)
Jev agreed with Qwen3 Max on >90% of verdicts. x.com/vargastartup/status/2101162776752538