Claude Sonnet 4.5
von Anthropic · Veroeffentlicht 2025-09-29
42.1
Durchschn. Score
$3.00/1M
Eingabepreis
$15.00/1M
Ausgabepreis
1.0M tokens (~500 books)
Kontextfenster
multimodal
Typ
Tested on 21 benchmarks with 42.1% average. Top scores: MATH level 5 (97.7%), OTIS Mock AIME 2024-2025 (77.8%), GPQA diamond (76.4%).
Benchmark-Ergebnisse
| Benchmark | Kategorie | Score | Bar |
|---|---|---|---|
| MATH level 5 | math | 97.7 | |
| OTIS Mock AIME 2024-2025 | math | 77.8 | |
| GPQA diamond | knowledge | 76.4 | |
| SWE-Bench verified | coding | 71.3 | |
| SWE-Bench Verified (Bash Only) | coding | 70.6 | |
| ARC-AGI | reasoning | 63.7 | |
| OSWorld | agentic | 62.9 | |
| Cybench | coding | 60.0 | |
| DeepResearch Bench | knowledge | 52.6 | |
| WeirdML | coding | 47.7 | |
| Terminal Bench | coding | 46.5 | |
| SimpleBench | reasoning | 45.2 | |
| SimpleQA Verified | knowledge | 23.6 | |
| FrontierMath-2025-02-28-Private | math | 15.2 | |
| GSO-Bench | coding | 14.7 | |
| ARC-AGI-2 | reasoning | 13.6 | |
| Chess Puzzles | knowledge | 12.0 | |
| PostTrainBench | knowledge | 9.9 | |
| VPCT | knowledge | 9.7 | |
| HLE | knowledge | 9.4 | |
| FrontierMath-Tier-4-2025-07-01-Private | math | 4.2 |
Aehnliche Modelle
Google DeepMind
42.2
Google DeepMind
42.2
Anthropic
42.3
Meta
42.5