Claude Opus 5 Benchmarks: What the Numbers Actually Show

DEV Community

Claude Opus 5 Benchmarks: What the Numbers Actually Show

Opus 5 posts 79.2 on SWE-bench Pro against Opus 4.8 at 69.2. What the launch ratios mean, and the benchmark row I refused to score.

Opus 5 posts 79.2 on SWE-bench Pro against Opus 4.8 at 69.2. What the launch ratios mean, and the benchmark row I refused to score.