Model Comparator
Compare up to four models using the same criteria.
2 models compared
- Gemma 4 31B (Reasoning)
Google
- DeepSeek V3.2 Speciale
DeepSeek
✓ Best published value among measured models. — Data unavailable. These metrics do not establish an overall ranking.
| Metrics | Gemma 4 31B (Reasoning) | DeepSeek V3.2 Speciale DeepSeek | Add a model · 2/4 |
|---|---|---|---|
| Intelligence | Best: 14.7 | 14.5 | |
| Coding | 43.4 | — | |
| Math | — | 96.7 | |
| Agentic | 4.2 | — | |
| Speed | 35.5 t/s | — | |
| TTFT | 1.03 s | — | |
| Context window | Best: 256K | 128K | |
| Reasoning | Yes | Yes | |
| Open weights | Yes | Yes | |
| Cost / task | $0 | — |
Intelligence
- Gemma 4 31B (Reasoning)
- Best: 14.7
- DeepSeek V3.2 Speciale
- 14.5
Coding
- Gemma 4 31B (Reasoning)
- 43.4
- DeepSeek V3.2 Speciale
- —
Math
- Gemma 4 31B (Reasoning)
- —
- DeepSeek V3.2 Speciale
- 96.7
Agentic
- Gemma 4 31B (Reasoning)
- 4.2
- DeepSeek V3.2 Speciale
- —
Speed
- Gemma 4 31B (Reasoning)
- 35.5 t/s
- DeepSeek V3.2 Speciale
- —
TTFT
- Gemma 4 31B (Reasoning)
- 1.03 s
- DeepSeek V3.2 Speciale
- —
Context window
- Gemma 4 31B (Reasoning)
- Best: 256K
- DeepSeek V3.2 Speciale
- 128K
Reasoning
- Gemma 4 31B (Reasoning)
- Yes
- DeepSeek V3.2 Speciale
- Yes
Open weights
- Gemma 4 31B (Reasoning)
- Yes
- DeepSeek V3.2 Speciale
- Yes
Cost / task
- Gemma 4 31B (Reasoning)
- $0
- DeepSeek V3.2 Speciale
- —
Each reasoning level remains a separate configuration.
Score and cost
Each line joins the reasoning levels of a compared model; the large point marks the configuration in the table. The top-right corner holds the best score for the lowest cost. Select another point to compare that level.
- Gemma 4 31B (Reasoning)14.7 · $0
Not plotted for lack of data on these axes: DeepSeek V3.2 Speciale.