DeepSeek V3.2 vs Qwen3 30B A3B Thinking 2507
Lado a lado. Cada métrica. Cada benchmark.
| Tipo | DeepSeek V3.2 | Qwen3 30B A3B Thinking 2507 |
|---|---|---|
| Provider |  DeepSeek |  Alibaba Qwen |
| pontuação média | 53.0 | 66.2 |
| Preço de entrada | $0.26 | $0.08 |
| Preço de saída | $0.38 | $0.40 |
| Janela de contexto | 164K tokens (~82 books) | 131K tokens (~66 books) |
| Lançado em | 2025-12-01 | 2025-08-28 |
| Código aberto | Open Source | Open Source |
Pontuações de benchmark
6 benchmarks · DeepSeek V3.2: 5, Qwen3 30B A3B Thinking 2507: 0
| Benchmark | Categoria | DeepSeek V3.2 | Qwen3 30B A3B Thinking 2507 |
|---|---|---|---|
| OpenCompass — AIME2025 | math | 93.0 | 86.8 |
| OpenCompass — GPQA-Diamond | knowledge | 84.6 | 70.1 |
| OpenCompass — HLE | knowledge | 23.2 | 11.7 |
| OpenCompass — IFEval | language | 89.7 | 89.7 |
| OpenCompass — LiveCodeBenchV6 | coding | 75.4 | 59.4 |
| OpenCompass — MMLU-Pro | knowledge | 85.8 | 79.5 |