Index
sid_y87VxEBBIA / OSWorld: 50.7
Verdictunverifiable95%
· 4/24/2026Inline sourcing: unverifiable
Our claim
entire record- Benchmark
- Hpb8OjdhT9
- Model
- Claude Haiku 4.5
- Score
- 50.7
- Unit
- percent
- Date
- October 15, 2025
- Notes
- OSWorld-Verified computer-use benchmark. Outperforms Sonnet 4 (42.2%) and far exceeds Sonnet 3.5 (14%). Reported by Anthropic at release.
Source evidence
0 src · 0 checksNo evidence on file.
Case № DrtKMMtM7gFiled 4/24/2026Confidence 95%