Gemini 3.1
Gemini 3.1 Pro (preview)
Google DeepMind · proprietary-api proprietary-api
| Parameters (B) | unverified |
|---|---|
| Architecture | unverified |
| Layers / KV heads / head dim | — / — / — |
| Context window | unverified |
| Modalities | in: text; out: text |
| GGUF | not indicated |
| Commercial use | yes |
Scores
| Benchmark | Value | Uncertainty | Run by | Source |
|---|---|---|---|---|
| GPQA Diamond | 94.4% | ±0.0163 | third-party | epoch-ai |
| SWE-bench Verified | 75.6% | ±0.0195 | third-party | epoch-ai |
| OTIS Mock AIME 2024-2025 | 95.6% | ±0.0310 | third-party | epoch-ai |
| FrontierMath | 36.9% | ±0.0280 | third-party | epoch-ai |
| FrontierMath Tier 4 | 16.7% | ±0.0540 | third-party | epoch-ai |
| SimpleQA Verified | 77.3% | ±0.0133 | third-party | epoch-ai |
| Terminal-Bench | 80.2% | ±0.0260 | third-party | epoch-ai |
| Terminal-Bench | 78.4% | ±0.0180 | third-party | epoch-ai |
| Terminal-Bench | 78.4% | ±0.0180 | third-party | epoch-ai |
| Terminal-Bench | 74.8% | ±0.0260 | third-party | epoch-ai |
| Terminal-Bench | 61.4% | ±0.0410 | third-party | epoch-ai |
Provenance
- identity, developer, release date: https://epoch.ai/benchmarks (retrieved 2026-08-10)