Gemini 2.5 Pro (Preview 03-25): benchmark results

March 25 Gemini 2.5 Pro preview snapshot. Provider: Google. Released 2025-03-25. Access: API.

Unified ELO 1659 ± 1, rank #111 of 1392 rated models, from 755 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
BALROG BabaIsAI (VLM)53.3Progress (%)100
BALROG NetHack (VLM)1.6Progress (%)100
BuzzBench (Humour)71.09Humour Score (0-100)100
CodeGolf Bench73.9C++ Best %ile (self-reported)100
EuroEval Faroese NLU - FoSent91.61Sentiment classification Score (%)100
KOFFVQA89.67Overall Score (%)100
KORGym - Mathematical and Logical0.93Score100
KORGym - Puzzle0.94Score100
KORGym - Strategic0.84Score100
LLM2014 Logic 2025-0384.27Median Score100
MEGA-Bench64.7Overall Score (%)100
MEGA-Bench - Application: Information Extraction72.6Average Score (%)100

Interactive version: theaggregate.ai/model?slug=gemini-2-5-pro-preview-03-25 · How It Works · Data refreshed daily, snapshot 2026-09-05.