GPT-5
von OpenAI · Veroeffentlicht 2025-08-07
54.4
Durchschn. Score
$1.25/1M
Eingabepreis
$10.00/1M
Ausgabepreis
400K tokens (~200 books)
Kontextfenster
multimodal
Typ
Tested on 26 benchmarks with 54.4% average. Top scores: MATH level 5 (98.1%), Fiction.LiveBench (97.2%), OTIS Mock AIME 2024-2025 (91.4%).
Benchmark-Ergebnisse
| Benchmark | Kategorie | Score | Bar |
|---|---|---|---|
| MATH level 5 | math | 98.1 | |
| Fiction.LiveBench | knowledge | 97.2 | |
| OTIS Mock AIME 2024-2025 | math | 91.4 | |
| Aider polyglot | coding | 88.0 | |
| Lech Mazur Writing | knowledge | 87.2 | |
| GPQA diamond | knowledge | 81.6 | |
| GeoBench | knowledge | 81.0 | |
| SWE-Bench verified | coding | 73.5 | |
| ARC-AGI | reasoning | 65.7 | |
| SWE-Bench Verified (Bash Only) | coding | 65.0 | |
| WeirdML | coding | 60.7 | |
| DeepResearch Bench | knowledge | 55.1 | |
| Professional Reasoning — Finance | knowledge | 51.3 | |
| SimpleQA Verified | knowledge | 50.6 | |
| Terminal Bench | coding | 49.6 | |
| VPCT | knowledge | 49.0 | |
| Professional Reasoning — Legal | knowledge | 49.0 | |
| SimpleBench | reasoning | 48.0 | |
| Chess Puzzles | knowledge | 37.0 | |
| Balrog | knowledge | 32.8 | |
| FrontierMath-2025-02-28-Private | math | 32.4 | |
| HLE | knowledge | 21.6 | |
| APEX-Agents | agentic | 18.3 | |
| FrontierMath-Tier-4-2025-07-01-Private | math | 12.5 | |
| ARC-AGI-2 | reasoning | 9.9 | |
| GSO-Bench | coding | 6.9 |
Aehnliche Modelle
Alibaba Qwen
54.4
OpenAI
54.0
xAI
54.8
OpenAI
55.0