o3 Mini vs o4 Mini
Lado a lado. Cada métrica. Cada benchmark.
| Tipo | o3 Mini | o4 Mini |
|---|---|---|
| Provider |  OpenAI | OpenAI |
| puntuación promedio | 38.4 | 53.2 |
| Precio de entrada | $1.10 | $1.10 |
| Precio de salida | $4.40 | $4.40 |
| Ventana de contexto | 200K tokens (~100 books) | 200K tokens (~100 books) |
| Publicado el | 2025-01-31 | 2025-04-16 |
| Código abierto | Proprietary | Proprietary |
Puntuaciones de benchmark
15 benchmarks · o3 Mini: 0, o4 Mini: 15
| Benchmark | Categoría | o3 Mini | o4 Mini |
|---|---|---|---|
| Aider polyglot | coding | 60.4 | 72.0 |
| ARC-AGI | reasoning | 34.5 | 58.7 |
| ARC-AGI-2 | reasoning | 3.0 | 6.1 |
| CadEval | coding | 54.0 | 62.0 |
| Chess Puzzles | knowledge | 17.0 | 26.0 |
| Fiction.LiveBench | knowledge | 50.0 | 77.8 |
| FrontierMath-2025-02-28-Private | math | 12.4 | 24.8 |
| FrontierMath-Tier-4-2025-07-01-Private | math | 4.2 | 6.3 |
| GPQA diamond | knowledge | 69.4 | 72.8 |
| GSO-Bench | coding | 1.3 | 3.6 |
| Lech Mazur Writing | knowledge | 61.7 | 75.0 |
| MATH level 5 | math | 96.5 | 97.8 |
| OTIS Mock AIME 2024-2025 | math | 76.9 | 81.7 |
| SimpleBench | reasoning | 7.4 | 26.4 |
| WeirdML | coding | 43.7 | 52.6 |