DeepSeek V3.2 Speciale vs Granite 4.0 H Small
DeepSeek vs IBM — benchmarks, pricing, and capabilities side by side.
- •DeepSeek V3.2 Speciale has the higher intelligence index (89.9 vs 35.7)
- •Granite 4.0 H Small is cheaper ($0.10 vs $0.29 per 1M input)
| DeepSeek V3.2 Speciale | Granite 4.0 H Small | |
|---|---|---|
| Intelligence index | 89.9 | 35.7 |
| Developer | DeepSeek | IBM |
| Type | LLM | LLM |
| Access | Open weights | — |
| Context window | 163,840 tokens | — |
| Input price | $0.29 / 1M | $0.10 / 1M |
| Output price | $0.43 / 1M | $0.30 / 1M |
| Speed | — | 524 tok/s |
| Released | December 1, 2025 | September 22, 2025 |
| Parameters | — | — |
| Input modalities | Text | — |
| Output modalities | Text | — |
Shared benchmarks
DeepSeek V3.2 Speciale
Granite 4.0 H Small
AIME 2025
96.7
13.7
GPQA Diamond
87.1
41.6
Humanity’s Last Exam
26.1
3.7
LiveCodeBench
89.6
25.1
MMLU-Pro
86.3
62.4