GLM 4.7 vs DeepSeek V3.2 Speciale
Lado a lado. Cada métrica. Cada benchmark.
| Tipo | GLM 4.7 | DeepSeek V3.2 Speciale |
|---|---|---|
| Provider |  z-ai |  DeepSeek |
| puntuación promedio | 50.5 | 78.2 |
| Precio de entrada | $0.39 | $0.40 |
| Precio de salida | $1.75 | $1.20 |
| Ventana de contexto | 203K tokens (~101 books) | 164K tokens (~82 books) |
| Publicado el | 2025-12-22 | 2025-12-01 |
| Código abierto | Open Source | Open Source |
Puntuaciones de benchmark
6 benchmarks · GLM 4.7: 2, DeepSeek V3.2 Speciale: 4
| Benchmark | Categoría | GLM 4.7 | DeepSeek V3.2 Speciale |
|---|---|---|---|
| OpenCompass — AIME2025 | math | 95.4 | 96.0 |
| OpenCompass — GPQA-Diamond | knowledge | 86.9 | 86.7 |
| OpenCompass — HLE | knowledge | 25.4 | 28.6 |
| OpenCompass — IFEval | language | 90.2 | 91.7 |
| OpenCompass — LiveCodeBenchV6 | coding | 83.8 | 80.9 |
| OpenCompass — MMLU-Pro | knowledge | 84.0 | 85.5 |