Claude Haiku 5.5
- Made by
- Anthropic
- Released
- Announced in
- Introducing Claude Haiku 5.5anthropic.com
- Event
- Claude Haiku 5.5
- API record
- /api/v1/models/claude-haiku-5-5
Benchmark results
| Benchmark | Score | Source check |
|---|---|---|
| Humanity's Last ExamNo tools | 45.9% | CheckedA person checked this number against its source on 16 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.CorroboratedThe same number is printed in 1 other lab document: 1 |
| Humanity's Last ExamWith tools | 57.4% | CheckedA person checked this number against its source on 19 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.CorroboratedThe same number is printed in 1 other lab document: 1 |
| SWE-bench Proadaptive thinking, max effort; average of 5 trials · The source doesn't say whether this is the public or the commercial set | 64.8% | CheckedA person checked this number against its source on 17 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced. |
| Terminal-Bench 4.0 | 39.2% | CheckedA person checked this number against its source on 17 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.CorroboratedThe same number is printed in 1 other lab document: 1 |
| Terminal-Bench Science 0.1Claude Code in --bare mode; max effort; no fallback model, so safeguard stops scored as failures; standard error ±3.7 points | 20.6% | CheckedA person checked this number against its source on 19 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced. |
| OSWorld 2.1Offline set, partial score · 82-task offline subset, no internet access; max effort; mean of 5 attempts | 72.4% | CorroboratedThe same number is printed in 1 other lab document: 1 |
| OSWorld 2.1Offline set, strict pass rate · 82-task offline subset, no internet access; max effort; mean of 5 attempts | 37.1% | 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. |
| FrontierCode 1.1Main set · max effort; run by Cognition in Claude Code; mean of 5 runs | 46.4% | CorroboratedThe same number is printed in 1 other lab document: 1 |
| xhigh effort; run by Cognition in Claude Code; mean of 5 runs · in Claude Haiku 5.5 System Card | 45.8% | 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. |
| FrontierCode 1.1Extended set · max effort; run by Cognition in Claude Code; mean of 5 runs | 58.4% | 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. |
Sources
Reported by Anthropic
- Introducing Claude Haiku 5.5anthropic.com · 7 Oct 2026
Our capture of the source - Claude Haiku 5.5 System Cardanthropic.com · 7 Oct 2026
Our capture of page 111 
Our capture of page 128