DeepSeek V3.2 vs Gemini 3.1 Pro Preview
Head to head on 24 shared benchmarks, with price, context window, and release dates.
Benchmark scores
57evals · sort by any column, ⤢ to expand fullscreen
| Model | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Gemini 3.1 Pro Preview Google (Alphabet Inc.) | 64.5% | 22.0% | 1489elo | 68.2 | 36.9% | 94.1% | 47.0% | 77.1% | 84.4% | 82.4% | 76.5% | 78.5% | 79.1% | 85.4% | 91.0% | 84.0% | 86.3% | 26.0% | 58.9% | 72.9% | 53.8% | 79.4 | 32.0% | 95.6% | - | - | - | 76.8% | 12.3% | 26.5% | - | 2 | 73.1% | 22.8% | 1142elo | 43.0% | 42.5% | 16.7% | 1314elo | 93.2% | 15.1% | - | 0.0% | - | 59.1% | 76.1% | - | 69.4% | 76.2% | 48.7% | 72.7% | - | - | 54.2% | 70.7% | 41.9% | 56.1% | 58.9% |
| DeepSeek V3.2 DeepSeek | 47.9% | 16.0% | 1511elo | 66.8 | 22.1% | 75.1% | 11.2% | 49.0% | 68.0% | 65.9% | 64.6% | 50.0% | 48.2% | 70.4% | 85.0% | 77.2% | 57.5% | 8.0% | 38.7% | 68.2% | 32.6% | 65 | 5.1% | 78.9% | 74.7% | 74.2% | 59.0% | - | - | - | 0.0% | - | - | - | - | - | - | - | - | - | - | 31.3% | - | 59.3% | - | - | 83.7% | - | - | - | - | 59.0% | 70.0% | - | - | - | - | 51.7% |
2 / 2 models
| Attribute | DeepSeek V3.2 DeepSeek | Gemini 3.1 Pro Preview Google (Alphabet Inc.) |
|---|---|---|
| Description | DeepSeek V3.2 is an AI model from DeepSeek, released with open weights. | Gemini 3.1 Pro Preview is an AI model from Google (Alphabet Inc.). |
| Family | Deepseek | Gemini |
| Params | - | - |
| Open weights | open | - |
| License | mit | - |
| Released | 1 Dec 2025 | 19 Feb 2026 |
| Context | 163,840 | 1,048,576 |
| Input $/1M | $0.28 | $2.00 |
| Output $/1M | $0.42 | $12.00 |
| Cache read $/1M | $0.130 | $0.200 |
| Cache write $/1M | - | $0.375 |
| Speed | 0 tok/s | 122 tok/s |
| Latency | 0.00 s | 20.68 s |
| Modalities | text | audio, file, image, text, video |
| Providers | - | - |
| Publisher | DeepSeek | Google (Alphabet Inc.) |
| Reported scores | Aider Polyglot Benchmark74.2% (Aider) MathArena57.47% GPQA Diamond75.1% (AA) Humanity's Last Exam (HLE)11.2% (AA) | GPQA Diamond94.1% (AA) Humanity's Last Exam (HLE)47.0% (AA) IFBench77.1% (AA) SciCode58.9% (AA) τ²-bench (Tau²-bench)95.6% (AA) |
FAQ
- Is DeepSeek V3.2 or Gemini 3.1 Pro Preview better?
- Across the 24 benchmarks both models report on Sophon, Gemini 3.1 Pro Preview leads on 23 of them. Which one is "better" depends on the benchmark - the table above shows every shared score.
- How do DeepSeek V3.2 and Gemini 3.1 Pro Preview compare on GPQA Diamond?
- DeepSeek V3.2 scores 75.1% and Gemini 3.1 Pro Preview scores 94.1% on GPQA Diamond.
- Which is cheaper, DeepSeek V3.2 or Gemini 3.1 Pro Preview?
- DeepSeek V3.2 is cheaper at $0.42 per million output tokens against $12.00 for Gemini 3.1 Pro Preview - about 28.6x.
- When were DeepSeek V3.2 and Gemini 3.1 Pro Preview released?
- DeepSeek V3.2 was released 1 Dec 2025 by DeepSeek. Gemini 3.1 Pro Preview was released 19 Feb 2026 by Google (Alphabet Inc.).
Comparing something else? Build your own side-by-side.