Our first Opus 5 results from running our custom SWE-bench on our Rails codebase: it tied Fable 5 on quality. Its average runtime per task was ~2 minutes shorter, and its average cost per task was ~$10 vs. ~$14 for Fable 5.
This continues a pattern we’ve seen over the last few…
Our first Opus 5 results from running our custom SWE-bench on our Rails codebase: it tied Fable 5 on quality. Its average runtime per task was ~2 minutes shorter, and its average cost per task was ~$10 vs. ~$14 for Fable 5.
This continues a pattern we’ve seen over the last few…
Our first Opus 5 results from running our custom SWE-bench on our Rails codebase: it tied Fable 5 on quality. Its average runtime per task was ~2 minutes shorter, and its average cost per task was ~$10 vs. ~$14 for Fable 5.
This continues a pattern we’ve seen over the last few…
Our first Opus 5 results from running our custom SWE-bench on our Rails codebase: it tied Fable 5 on quality. Its average runtime per task was ~2 minutes shorter, and its average cost per task was ~$10 vs. ~$14 for Fable 5.
This continues a pattern we’ve seen over the last few…