DeepSeek V3.2 vs Qwen3 32B
Lado a lado. Cada métrica. Cada benchmark.
| Tipo | DeepSeek V3.2 | Qwen3 32B |
|---|---|---|
| Provider |  DeepSeek |  Alibaba Qwen |
| puntuación promedio | 53.0 | 58.2 |
| Precio de entrada | $0.26 | $0.08 |
| Precio de salida | $0.38 | $0.24 |
| Ventana de contexto | 164K tokens (~82 books) | 41K tokens (~20 books) |
| Publicado el | 2025-12-01 | 2025-04-28 |
| Código abierto | Open Source | Open Source |
Puntuaciones de benchmark
8 benchmarks · DeepSeek V3.2: 8, Qwen3 32B: 0
| Benchmark | Categoría | DeepSeek V3.2 | Qwen3 32B |
|---|---|---|---|
| Aider polyglot | coding | 74.2 | 40.0 |
| Chatbot Arena Elo — Overall | arena | 1424.4 | 1347.0 |
| OpenCompass — AIME2025 | math | 93.0 | 70.3 |
| OpenCompass — GPQA-Diamond | knowledge | 84.6 | 67.3 |
| OpenCompass — HLE | knowledge | 23.2 | 8.5 |
| OpenCompass — IFEval | language | 89.7 | 86.0 |
| OpenCompass — LiveCodeBenchV6 | coding | 75.4 | 57.6 |
| OpenCompass — MMLU-Pro | knowledge | 85.8 | 78.0 |