Category scores
Out of 100: 100 = the best tracked model in the category.
| Reasoning | 46 | |
| Mathematics | 60 | |
| Code | 67 | |
| Agents & real work | 72 | |
| Knowledge | 23 | |
| Long context | 93 | |
| Multimodal | — |
EditionUpdated daily
Ranking / Claude Haiku 5.5
Claude Haiku 5.5
Anthropic · Proprietary · context 1M. Ranked #23 of 26 with 544 ERAZ points out of 1000.
ERAZ rank#23/ 26
Points544/ 1000
Coverage70%
Benchmarks won0/ 14
Input price / M$0.1
Output price / M$0.5
Points over time
| Day | Points | Change |
|---|
Results on each benchmark
| Benchmark | Score | Rank | Out of 100 | Source |
|---|---|---|---|---|
| Humanity's Last Exam | 44.4% | #16 / 26 | 72 | src ↗ |
| ARC-AGI-2 | 61.3% | #12 / 13 | 65 | src ↗ |
| ARC-AGI-3 | 0.5% | #9 / 9 | 1 | src ↗ |
| FrontierMath (Tiers 1-3) | 75.1% | #10 / 16 | 80 | src ↗ |
| FrontierMath (Tier 4) | 46.3% | #10 / 16 | 46 | src ↗ |
| SWE-bench Pro | not published | — | — | |
| FrontierSWE v2 | 43.8% | #7 / 14 | 67 | src ↗ |
| Terminal-Bench 2.1 | not published | — | — | |
| OSWorld 2.0 | not published | — | — | |
| GDPval-AA | 1620 | #15 / 23 | 72 | src ↗ |
| GPQA Diamond | not published | — | — | |
| AA-Omniscience | 10.7 | #18 / 26 | 23 | src ↗ |
| AA-LCR | 82.7% | #13 / 26 | 93 | src ↗ |
| MMMU-Pro | not published | — | — |
Artificial Analysis Intelligence Index : 43.4 · Source page ↗
Nearby models in the ranking
</> Embed this ranking
Show Claude Haiku 5.5's ranking on your site or in a README. The badge updates every day.
