Gemini 3.5 Flash: benchmark results
Google's speed-oriented Gemini 3.5 Flash model, balancing capability with lower latency. Provider: Google. Released 2026-05-19. Access: API.
Unified ELO 1726 ± 1, rank #28 of 1392 rated models, from 268 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| APEX-Agents | 66.1 | Mean Score (ReAct) (self-reported) | 100 |
| Arabic Broad Leaderboard | 9.25 | Average Score (0-10) | 100 |
| CAIS Vision Capabilities Index | 65.9 | Vision Capabilities Index (self-reported) | 100 |
| LiveMathematicianBench | 49.5 | Accuracy (%) | 100 |
| PerceptionBench - Hallucination | 50.6 | Perception Hallucination (%) | 100 |
| SciCode-Verified | 97.2 | Subproblem Resolve Rate (%) | 100 |
| Nejumi 4 - ALT - Ethics & Morality | 99 | Score (%) | 99.6 |
| Nejumi 4 - GLP - General Knowledge | 85.27 | Score (%) | 99.2 |
| Roboflow Vision Evals - Visual Understanding | 79.1 | Pass Rate (%) | 98.7 |
| AI for Education Visual Reasoning | 86 | Accuracy (%) | 98.5 |
| AI for Education Visual Reasoning - match (figure) | 85.2 | Accuracy (%) | 98.5 |
| AI for Education Visual Reasoning - reasoning by analogy | 88.8 | Accuracy (%) | 98.5 |
Interactive version: theaggregate.ai/model?slug=gemini-3-5-flash · How It Works · Data refreshed daily, snapshot 2026-09-05.