Gemini 3.1 Flash Lite (High): benchmark results
Gemini 3.1 Flash Lite evaluated at the high reasoning-effort setting. Provider: Google. Released 2026-05-07. Access: API.
Unified ELO 1642 ± 1, rank #379 of 3081 rated models, from 41 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Korean CSAT 2026 (Easy Mode) - Korean | 100 | Points (out of 100) | 92.5 |
| SLR-Bench - Hard | 18 | Accuracy (%) | 88.2 |
| SLR-Bench - Easy | 96 | Accuracy (%) | 87.3 |
| SLR-Bench - Medium | 48 | Accuracy (%) | 87.3 |
| ObviousBench | 97.92 | Answer pass³ (%) | 80.9 |
| Korean CSAT 2026 (Easy Mode) - Chemistry I | 48 | Points (out of 50) | 79.9 |
| Sonar LLM Leaderboard - Java - Pass Rate | 79.56 | Passing tests (%) | 69.4 |
| Sonar LLM Leaderboard - Java - Bug Density | 0.73 | Bugs per 1,000 lines of code (lower is better) | 66.7 |
| Korean CSAT 2026 (Easy Mode) - Total | 419.5 | Points (out of 450) | 64.6 |
| Epoch AI - GPQA Diamond | 81.82 | Accuracy (%) | 64.1 |
| OTIS Mock AIME 2024-25 | 80 | Accuracy (%) | 63.1 |
| Korean CSAT 2026 (Easy Mode) - Physics I | 35 | Points (out of 50) | 62.9 |
Interactive version: theaggregate.ai/model?slug=gemini-3-1-flash-lite-high · How It Works · Data refreshed daily, snapshot 2026-09-21.