Gemini 2.5 Flash (Preview) (Reasoning) — benchmark results

Provider: Google. Released 2025-06-17. Access: API.

Unified ELO 1721 ± 62, rank #240 of 1841 rated models, from 7 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
AA MATH-50098.07Accuracy (%)92.2
AA MMLU-Pro79.98Accuracy (%)66.9
AA Humanity's Last Exam11.56Accuracy (%)65.9
AA LiveCodeBench50.48Pass@1 (%)56.4
Epoch AI - Scicode35.88Score55.4
Artificial Analysis Intelligence Index17.55Intelligence Index53.6
AA GPQA Diamond69.8Accuracy (%)52.3

Interactive version: theaggregate.ai/model?slug=gemini-2-5-flash-preview-reasoning · How It Works · Data refreshed daily, snapshot 2026-07-25.