GPT-5.4 vs Gemini 3 Pro
나란히 비교. 모든 지표. 모든 벤치마크.
| 유형 | GPT-5.4 | Gemini 3 Pro |
|---|---|---|
| Provider |  OpenAI |  Google DeepMind |
| 평균 점수 | 59.0 | 60.5 |
| 입력 가격 | $2.50 | - |
| 출력 가격 | $15.00 | - |
| 컨텍스트 윈도우 | 1.1M tokens (~525 books) | - |
| 출시일 | 2026-03-05 | 2024-01-01 |
| 오픈소스 | Proprietary | Proprietary |
벤치마크 점수
16 benchmarks · GPT-5.4: 13, Gemini 3 Pro: 3
| 벤치마크 | 카테고리 | GPT-5.4 | Gemini 3 Pro |
|---|---|---|---|
| APEX-Agents | agentic | 35.9 | 18.4 |
| ARC-AGI | reasoning | 93.7 | 75.0 |
| ARC-AGI-2 | reasoning | 74.0 | 31.1 |
| Artificial Analysis — Agentic Index | speed | 69.4 | 45.0 |
| Artificial Analysis — Coding Index | speed | 57.3 | 39.4 |
| Artificial Analysis — Quality Index | speed | 57.2 | 41.3 |
| Chatbot Arena Elo — Overall | arena | 1465.8 | 1486.2 |
| Chess Puzzles | knowledge | 44.0 | 31.0 |
| FrontierMath-2025-02-28-Private | math | 47.6 | 37.6 |
| FrontierMath-Tier-4-2025-07-01-Private | math | 27.1 | 18.8 |
| GPQA diamond | knowledge | 91.1 | 90.2 |
| OTIS Mock AIME 2024-2025 | math | 95.3 | 91.4 |
| PostTrainBench | knowledge | 20.2 | 18.1 |
| SimpleQA Verified | knowledge | 44.8 | 72.9 |
| SWE-Bench verified | coding | 76.9 | 72.9 |
| WeirdML | coding | 57.4 | 69.9 |