GPT-5 Mini vs DeepSeek V3.2
Côte à côte. Chaque métrique. Chaque benchmark.
| Type | GPT-5 Mini | DeepSeek V3.2 |
|---|---|---|
| Provider |  OpenAI |  DeepSeek |
| score moyen | 56.0 | 53.0 |
| Prix d'entrée | $0.25 | $0.26 |
| Prix de sortie | $2.00 | $0.38 |
| Fenêtre de contexte | 400K tokens (~200 books) | 164K tokens (~82 books) |
| Sorti le | 2025-08-07 | 2025-12-01 |
| Code source ouvert | Proprietary | Open Source |
Scores de benchmark
16 benchmarks · GPT-5 Mini: 10, DeepSeek V3.2: 6
| Benchmark | Catégorie | GPT-5 Mini | DeepSeek V3.2 |
|---|---|---|---|
| ARC-AGI | reasoning | 54.3 | 57.0 |
| ARC-AGI-2 | reasoning | 4.4 | 4.0 |
| FrontierMath-2025-02-28-Private | math | 27.2 | 22.1 |
| FrontierMath-Tier-4-2025-07-01-Private | math | 6.3 | 2.1 |
| GPQA diamond | knowledge | 66.7 | 77.9 |
| LiveBench — Agentic Coding | coding | 35.0 | 46.7 |
| LiveBench — Coding | coding | 76.1 | 75.7 |
| LiveBench — Data Analysis | reasoning | 49.6 | 45.0 |
| LiveBench — If | language | 64.2 | 23.1 |
| LiveBench — Language | language | 69.2 | 64.2 |
| LiveBench — Mathematics | math | 74.4 | 64.0 |
| LiveBench — Overall | knowledge | 61.0 | 51.8 |
| LiveBench — Reasoning | reasoning | 58.6 | 44.3 |
| OTIS Mock AIME 2024-2025 | math | 86.7 | 87.8 |
| SimpleQA Verified | knowledge | 21.0 | 27.5 |
| Terminal Bench | coding | 34.8 | 39.6 |