Gemini 2.5 Pro: benchmark results

Google Gemini 2.5 Pro model for advanced reasoning and multimodal tasks. Provider: Google. Released 2025-03-25. Access: API.

Unified ELO 1661 ± 1, rank #109 of 1392 rated models, from 1369 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
AI for Education Visual Maths - Algebra100Accuracy (%)100
Active Evidence-Seeking and Diagnostic Reasoni49.4Task 2: Active Seeking Exact Accuracy (self-reported)100
BIRD-Interact (c-Interact)20.92Normalized Reward100
BesiegeField - Car34.34Best-scaffold Mean Score100
BesiegeField - Catapult9.83Best-scaffold Mean Score100
Chartographer94.7ChartQA OA (self-reported)100
DeepResearch Bench - Effective Citations165.34Effective Citations100
Episodic Memory96.8Simple Recall (%)100
EuroEval Czech70.02Average Score (%)100
EuroEval Estonian62.38Average Score (%)100
EuroEval Hungarian67.51Average Score (%)100
EuroEval Norwegian Knowledge87.61Knowledge Average Score (%)100

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