Opus 5.5 scores worse on high effort than medium, SlopCodeBench finds
SlopCodeBench 10/2026 – Opus 5.5, Astra wins overall, xhigh considered harmful
On the full SlopCodeBench—36 problems and 196 checkpoints—GPT-6 Astra XHigh leads strict passes at 16.33%, just two checkpoints ahead of GPT-6.1 SOL Medium, which costs far less. Raising effort to XHigh cost 2.11–4.22x more, yet Opus 5.5 did better on medium (28 strict passes) than XHigh (25). Astra's final code still failed 1,435 test cases, and Opus Medium left the most slop-heavy code.
While Opus 5.5 Medium had more strict passes, Opus 5.5 XHigh wrote code that was close to ~5x cleaner.