Index
sid_y87VxEBBIA / SWE-bench Verified: 73.3
Verdictconfirmed98%
· 4/24/2026Inline sourcing: confirmed
Our claim
entire record- Benchmark
- WOSlsBTTmV
- Model
- Claude Haiku 4.5
- Score
- 73.3
- Unit
- percent
- Date
- October 15, 2025
- Notes
- Score reported by Anthropic averaged over 50 runs. One of the highest scores in its price tier. Tested via Anthropic's agent scaffold.
Source evidence
0 src · 0 checksNo evidence on file.
Case № 9iqsCKcMggFiled 4/24/2026Confidence 98%