DeepSeek V3.2 Speciale vs Gemma 3 4B Instruct
DeepSeek vs Google — benchmarks, pricing, and capabilities side by side.
- •DeepSeek V3.2 Speciale has the higher intelligence index (89.9 vs 31.7)
- •Gemma 3 4B Instruct is cheaper ($0.00 vs $0.29 per 1M input)
| DeepSeek V3.2 Speciale | Gemma 3 4B Instruct | |
|---|---|---|
| Intelligence index | 89.9 | 31.7 |
| Developer | DeepSeek | |
| Type | LLM | LLM |
| Access | Open weights | — |
| Context window | 163,840 tokens | — |
| Input price | $0.29 / 1M | $0.00 / 1M |
| Output price | $0.43 / 1M | $0.10 / 1M |
| Speed | — | — |
| Released | December 1, 2025 | March 12, 2025 |
| Parameters | — | — |
| Input modalities | Text | — |
| Output modalities | Text | — |
Shared benchmarks
DeepSeek V3.2 Speciale
Gemma 3 4B Instruct
AIME 2025
96.7
12.7
GPQA Diamond
87.1
29.1
Humanity’s Last Exam
26.1
5.2
LiveCodeBench
89.6
11.2
MMLU-Pro
86.3
41.7