Gemini 2.0 Pro 02-05 preview — benchmark results

Provider: Google. Released 2025-02-05. Access: API.

Unified ELO 1578 ± 63, rank #598 of 1839 rated models, from 7 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
HELM Safety BBQ95.6BBQ accuracy (%)72.1
HELM AIR-Bench68.4Refusal Rate (%)55.8
HELM Safety Anthropic Red Team98.7LM Evaluated Safety score (%)45.3
HELM Safety XSTest95.4LM Evaluated Safety score (%)44.8
HELM Safety90.5Mean score (self-reported)42.8
HELM Safety HarmBench65.3LM Evaluated Safety score (%)34.9
HELM Safety SimpleSafetyTests97.5LM Evaluated Safety score (%)32.6

Interactive version: theaggregate.ai/model?slug=gemini-2-0-pro-02-05-preview · How It Works · Data refreshed daily, snapshot 2026-08-05.