Frontier Log

Benchmark · Active

OSWorld 2.1

0204060Oct 202653.2%Opus 5.537.1%Haiku 5.50204060Oct 202653.2%Opus 5.537.1%Haiku 5.5

Results

ModelScoreSourceSource check
Claude Haiku 5.57 Oct 2026 · Anthropic37.1%Claude Haiku 5.5 System Card82-task offline subset, no internet access; max effort; mean of 5 attempts
Source checkedA person checked this number against its source on 20 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
Claude Sonnet 5.528 Sep 2026 · Anthropic48.8%Claude Haiku 5.5 System Card82-task offline subset, no internet access; max effort; mean of 5 attempts; reported later
Source checkedA person checked this number against its source on 20 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
Claude Opus 5.522 Sep 2026 · Anthropic53.2%Claude Haiku 5.5 System Card82-task offline subset, no internet access; max effort; mean of 5 attempts; reported later
Source checkedA person checked this number against its source on 20 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.

Benchmark

github.comHigher is better

XLANG LabIntroduced 16 Sep 2026

All benchmarks