Re:SWE-1.7 Reach Near GPT 5.5 and Opus Intelligence
by spate141 on Wednesday July 08, 2026 @01:10PM
(#48834484)
Feels like they discovers that if you build your own benchmark, you can win it
[ Parent
| on HN ]
Re:SWE-1.7 Reach Near GPT 5.5 and Opus Intelligence
by londons_explore on Wednesday July 08, 2026 @01:17PM
(#48834566)
Pretty sure most benchmarks are being gamed by people training on the test set deliberately or accidentally anyway.