08:15
2026-07-13
github.com
artificial-intelligence
Tell HN: One SWE-bench-Live task: $47 Opus failed, $1.46 GPT-5.6 passed
Seven AI models attempted a SWE-bench-Live task with internet and git history disabled; three passed and four failed. GPT-5.6 Luna solved the task for $1.46 while Opus 4.8 failed at $47.18, with cache…