Gemini 2.5 Pro (Preview 06-05): benchmark results

June 5 Gemini 2.5 Pro preview snapshot. Provider: Google. Released 2025-06-05. Access: API.

Unified ELO 1684 ± 1, rank #78 of 1392 rated models, from 41 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
LLM Stats (Vibe-Eval)67.2Score (%)100
SEAL - VISTA54.63Score98.4
SEAL - Fortress61.68Score96.4
Web-Bench25.3Pass@1 (%)95.7
Balrog43.3Score (self-reported)93.3
SEAL - TutorBench55.65Score92.3
SpeechMap Compliance89% Requests Completed89.4
CadEval64Score (self-reported)88.9
Context Arena MRCR (2-needle)77.5AUC@1M (%)88.8
Wolfram LLM Benchmarking Project59.1Correct Functionality (%)87.3
Fiction.LiveBench87.5Accuracy (%)87.2
LiveCodeBench84.3Pass@1 avg (%)85.2

Interactive version: theaggregate.ai/model?slug=gemini-2-5-pro-preview-06-05 · How It Works · Data refreshed daily, snapshot 2026-09-05.