Gemini 3.8 Flash: benchmark results

Provider: Google. Released 2026-09-02. Access: API.

Unified ELO 1778 ± 1, rank #6 of 1392 rated models, from 37 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
BioMysteryBench Human-Difficult56.5Accuracy (self-reported)100
LLM Stats (Terminal-Bench 2.1)89.4Score (%)100
LVBench87.8Score (self-reported)100
AI Chess Leaderboard (Continuation)1772Elo99.2
AI Chess Leaderboard (Reasoning)1849Elo99.1
Design Arena (UI Components)1368Elo99
BenchmarkList ECI152.3Capability Index (ECI)98.4
SimpleBench82.4Score (AVG@5)98
BenchLM78.4Overall Score97.5
SvelteBench100Average pass@1 (%)96.9
LLM Stats Score50.97LLM Stats Score (conservative rating)96
Design Arena (Game Dev)1344Elo95

Interactive version: theaggregate.ai/model?slug=gemini-3-8-flash · How It Works · Data refreshed daily, snapshot 2026-09-05.