DeepSeek V3.2 Speciale vs GLM 5
Lado a lado. Cada métrica. Cada benchmark.
| Tipo | DeepSeek V3.2 Speciale | GLM 5 |
|---|---|---|
| Provider |  DeepSeek |  z-ai |
| puntuación promedio | 78.2 | 57.6 |
| Precio de entrada | $0.40 | $0.72 |
| Precio de salida | $1.20 | $2.30 |
| Ventana de contexto | 164K tokens (~82 books) | 80K tokens (~40 books) |
| Publicado el | 2025-12-01 | 2026-02-11 |
| Código abierto | Open Source | Open Source |
Puntuaciones de benchmark
6 benchmarks · DeepSeek V3.2 Speciale: 4, GLM 5: 2
| Benchmark | Categoría | DeepSeek V3.2 Speciale | GLM 5 |
|---|---|---|---|
| OpenCompass — AIME2025 | math | 96.0 | 95.8 |
| OpenCompass — GPQA-Diamond | knowledge | 86.7 | 85.3 |
| OpenCompass — HLE | knowledge | 28.6 | 28.1 |
| OpenCompass — IFEval | language | 91.7 | 93.2 |
| OpenCompass — LiveCodeBenchV6 | coding | 80.9 | 86.2 |
| OpenCompass — MMLU-Pro | knowledge | 85.5 | 85.2 |