Gemini 2.0 Flash
제공 Google DeepMind · 출시일 2025-02-05
48.0
평균 점수
$0.10/1M
입력 가격
$0.40/1M
출력 가격
1.0M tokens (~524 books)
컨텍스트 윈도우
multimodal
유형
Tested on 20 benchmarks with 48.0% average. Top scores: Chatbot Arena Elo — Overall (1360.0%), HELM — IFEval (84.1%), MATH level 5 (82.2%).
벤치마크 점수
| 벤치마크 | 카테고리 | 점수 | Bar |
|---|---|---|---|
| Chatbot Arena Elo — Overall | arena | 1360.0 | |
| HELM — IFEval | language | 84.1 | |
| MATH level 5 | math | 82.2 | |
| HELM — WildBench | reasoning | 80.0 | |
| GeoBench | knowledge | 77.0 | |
| HELM — MMLU-Pro | knowledge | 73.7 | |
| MMLU | knowledge | 72.9 | |
| Lech Mazur Writing | knowledge | 71.5 | |
| Fiction.LiveBench | knowledge | 61.1 | |
| HELM — GPQA | knowledge | 55.6 | |
| GPQA diamond | knowledge | 52.2 | |
| HELM — Omni-MATH | math | 45.9 | |
| Aider polyglot | coding | 38.2 | |
| OTIS Mock AIME 2024-2025 | math | 31.0 | |
| CadEval | coding | 30.0 | |
| WeirdML | coding | 25.8 | |
| SimpleBench | reasoning | 17.3 | |
| The Agent Company | agentic | 11.4 | |
| FrontierMath-2025-02-28-Private | math | 1.7 | |
| ARC-AGI-2 | reasoning | 1.3 |