Claude Mythos Preview
제공 Anthropic · 출시일 2026-04-07
81.8
평균 점수
N/A
입력 가격
N/A
출력 가격
1.0M tokens (~500 books)
컨텍스트 윈도우
text
유형
Tested on 14 benchmarks with 81.8% average. Top scores: USAMO (97.6%), GPQA diamond (94.5%), SWE-Bench verified (93.9%).
벤치마크 점수
| 벤치마크 | 카테고리 | 점수 | Bar |
|---|---|---|---|
| USAMO | math | 97.6 | |
| GPQA diamond | knowledge | 94.5 | |
| SWE-Bench verified | coding | 93.9 | |
| CharXiv Reasoning (with tools) | reasoning | 93.2 | |
| MMMLU | knowledge | 92.7 | |
| SWE-bench Multilingual | coding | 87.3 | |
| CharXiv Reasoning | reasoning | 86.1 | |
| Terminal Bench | coding | 82.0 | |
| GraphWalks BFS 256K-1M | reasoning | 80.0 | |
| OSWorld | agentic | 79.6 | |
| SWE-bench Pro | coding | 77.8 | |
| HLE (with tools) | reasoning | 64.7 | |
| SWE-bench Multimodal | coding | 59.0 | |
| HLE | knowledge | 56.8 |