Gemini 3.1 Flash Lite: benchmark results
Google's fastest, most cost-efficient Gemini 3.1 tier for low-latency, high-volume workloads. Provider: Google. Released 2026-05-07. Access: API.
Aggregate Rating 1702 ± 4, rank #191 of 1656 rated models, from 978 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| 3D-DefectBench - Texture (Expert Labels) | 0.41 | Macro MCC (Matthews correlation from -1 to 1, macro-averaged | 100 |
| AfriScience-MT - Document Level (1-Shot) | 48.3 | SSA-COMET (0-100; whole-summary translation, mean over the 1 | 100 |
| Arabic Broad Leaderboard - MMLU | 10 | Average Score (0-10) | 100 |
| CoMMET (Image-Dependent) - Intention | 29.27 | Story accuracy on the 41 intention StoryTurns among the 192 | 100 |
| Colon-Bench - Lesion Classification F1 | 80.7 | F1 (%) of the lesion-present class when classifying Colon-Be | 100 |
| EgoArgus (Assistance Decision) | 81.17 | Assistance-decision F1 (%; whether help is required, from th | 100 |
| EgoArgus (Assistance Decision) - Multimodal Grounded | 93.95 | Assistance-decision F1 (%; whether help is required, from th | 100 |
| EgoArgus (Assistance Decision) - Text-Grounded | 94.63 | Assistance-decision F1 (%; whether help is required, from th | 100 |
| EgoArgus (Assistance Decision) - Video-Grounded Off-Topic | 74.31 | Assistance-decision F1 (%; whether help is required, from th | 100 |
| Evals for Every Language - Catalan (ca) | 80.38 | Average Score (%) | 100 |
| Evals for Every Language - Greek (el) | 79.42 | Average Score (%) | 100 |
| Evals for Every Language - English (en) | 89 | Average Score (%) | 100 |
Interactive version: theaggregate.ai/model?slug=gemini-3-1-flash-lite · How It Works · Data refreshed daily, snapshot 2026-10-20.