Gemini 2.0 Pro (Preview 02-05): benchmark results
February 5 Gemini 2.0 Pro preview snapshot. Provider: Google. Released 2025-02-05. Access: API.
Aggregate Rating 1653 ± 13, rank #307 of 1656 rated models, from 607 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| HELM Audio - Air Bench Chat | 3.25 | GPT-4o Judge Critique | 100 |
| HELM Audio - Common Voice 15 | 9.21 | WER | 100 |
| HELM VHELM - Pope | 89.6 | PEM | 100 |
| OpenCompass Knowledge - Engineering - English (CompassBench 2501) | 96 | Score (%) | 100 |
| OpenCompass Knowledge - Humanities - Chinese (CompassBench 2501) | 96 | Score (%) | 100 |
| OpenCompass LLM - Knowledge (CompassBench 2501) | 92 | Score (%) | 100 |
| OpenCompass LLM - Language - Chinese (CompassBench 2501) | 75.1 | Score (%) | 100 |
| SuperCLUE General (March 2025) - Science Reasoning | 70.71 | Score | 100 |
| OpenCompass Knowledge - Humanities - English (CompassBench 2501) | 89 | Score (%) | 98.7 |
| HELM VHELM - Math Vista - High School (Free Form) | 96.77 | Exact Match (%) | 97.9 |
| HELM VHELM - Mmmu - Art | 93.33 | Prefix Quasi-Exact Match (%) | 97.9 |
| HELM VHELM - Robustness | 82.98 | Mean win rate | 97.9 |
Interactive version: theaggregate.ai/model?slug=gemini-2-0-pro-preview-02-05 · How It Works · Data refreshed daily, snapshot 2026-10-20.