Google · model
Gemini 3.1 Flash-Lite benchmark scores
As of 2026-11-22, Gemini 3.1 Flash-Lite (Google) has measured scores on 14 benchmarks and estimated scores on 105 more.
| Benchmark | Score | Source |
|---|---|---|
| AA-GPQA Diamond | 82.1% | estimated ± 3.6 pp, high confidence |
| AA-HLE | 19.4% | estimated ± 6.2 pp, medium confidence |
| AA-SciCode | 45.2% | estimated ± 3.8 pp, medium confidence |
| AIME26 | 94.3% | estimated ± 1.4 pp, high confidence |
| Artificial Analysis Intelligence Index | 19.1% | estimated ± 5.8 pp, medium confidence |
| CritPt | 1.8% | estimated ± 5.3 pp, medium confidence |
| GDPval-AA | 18.8% | estimated ± 10.5 pp, low confidence |
| GPQA Diamond | 85.7% | estimated ± 1.8 pp, medium confidence |
| HLE | 40.4% | estimated ± 9.2 pp, low confidence |
| LiveCodeBench v6 | 86.5% | estimated ± 2.2 pp, medium confidence |
| AA-Omniscience Accuracy | 37.0% | estimated ± 6.5 pp, medium confidence |
| SciCode | 44.0% | estimated ± 4.5 pp, low confidence |
| Terminal-Bench 2.1 | 16.8% | estimated ± 6.7 pp, low confidence |
| BrowseComp | 67.1% | estimated ± 8.8 pp, medium confidence |
| HLE w/ tools | 45.6% | estimated ± 5.7 pp, medium confidence |
| AA Coding Index | 41.7% | estimated ± 5.6 pp, medium confidence |
| AA-MMMU-Pro | 71.9% | estimated ± 4.9 pp, medium confidence |
| APEX-Agents-AA | 12.1% | estimated ± 9.3 pp, medium confidence |
| SWE-bench Verified | 73.0% | estimated ± 4.4 pp, high confidence |
| CyberGym | 48.7% | estimated ± 11.0 pp, low confidence |
| JobBench | 18.7% | estimated ± 14.2 pp, low confidence |
| SWE-bench Pro | 53.0% | estimated ± 2.9 pp, low confidence |
| WideResearch | 67.6% | estimated ± 3.3 pp, high confidence |
| JevBench 1.4 | 14.3% | measured |
| AA Agentic Index | 15.3% | estimated ± 9.4 pp, medium confidence |
| MMLU-Pro | 85.6% | estimated ± 4.8 pp, high confidence |
| FailureSensorIQ | 62.0% | estimated ± 1.7 pp, medium confidence |
| GPQA | 83.3% | estimated ± 4.5 pp, high confidence |
| MMLU | 93.5% | estimated ± 1.2 pp, medium confidence |
| Vals GPQA Diamond | 81.1% | measured |
| Vals LiveCodeBench | 80.1% | measured |
| Vals MMLU-Pro | 86.2% | measured |
| MMMU | 80.0% | estimated ± 3.7 pp, medium confidence |
| Chess Puzzles | 19.3% | estimated ± 12.3 pp, low confidence |
| AA MMLU-Pro | 87.3% | estimated ± 1.1 pp, high confidence |
| Gert Labs | 25.7% | measured |
| AA Terminal-Bench 2.1 | 79.2% | estimated ± 2.5 pp, medium confidence |
| ApprenticeBench | 3.2% | estimated ± 10.4 pp, low confidence |
| ARC-AGI-1 | 70.1% | estimated ± 6.5 pp, medium confidence |
| ARC-AGI-2 | 20.2% | estimated ± 13.8 pp, low confidence |
| CursorBench 3.1 | 36.5% | estimated ± 5.8 pp, low confidence |
| CursorBench 3.2 | 48.6% | estimated ± 5.4 pp, low confidence |
| EBR-bench | 0.0% | estimated ± 9.1 pp, low confidence |
| FrontierCode 1.1 Main | 14.8% | estimated ± 5.2 pp, low confidence |
| FrontierMath v2 (Tier 4) | 2.3% | estimated ± 4.0 pp, high confidence |
| FrontierMath v2 (Tiers 1-3) | 27.7% | measured |
| FrontierSWE v2 | 4.7% | estimated ± 9.5 pp, low confidence |
| MazeBench | 0.0% | estimated ± 2.7 pp, high confidence |
| Mystery Game Puzzles | 11.8% | estimated ± 9.8 pp, low confidence |
| NYT Connections | 8.2% | measured |
| OfficeQA Pro | 52.3% | estimated ± 7.0 pp, low confidence |
| OSWorld-Verified | 63.5% | estimated ± 8.6 pp, low confidence |
| ParseBench-1 | 56.7% | measured |
| Terminal-Bench 4.1 | 4.0% | estimated ± 14.9 pp, low confidence |
| Terminal-Bench Hard | 43.5% | estimated ± 3.9 pp, medium confidence |
| Vals SWE-bench | 62.8% | measured |
| Terminal-Bench 2.1 (Vals AI) | 34.1% | measured |
| VulcanBench v3 | 80.9% | estimated ± 5.6 pp, low confidence |
| CWE-Bench-AA | 36.5% | estimated ± 8.1 pp, low confidence |
| AA Cyber Index | 11.4% | estimated ± 12.4 pp, low confidence |
| DeepsecBench-AA | 7.7% | estimated ± 8.2 pp, low confidence |
| AA ITBench | 36.1% | estimated ± 4.6 pp, low confidence |
| CursorBench 4.0 | 30.7% | estimated ± 3.4 pp, medium confidence |
| CWE-bench v1 | 36.6% | estimated ± 5.7 pp, low confidence |
| DeepSWE | 56.3% | estimated ± 7.2 pp, low confidence |
| HLE w/o tools | 34.4% | estimated ± 4.9 pp, high confidence |
| OSWorld 2.1 | 5.0% | estimated ± 3.6 pp, medium confidence |
| PostTrainBench v1.1 | 9.6% | estimated ± 3.7 pp, medium confidence |
| StealthMark | 41.3% | estimated ± 10.7 pp, low confidence |
| SWE-bench Multimodal | 3.3% | estimated ± 5.6 pp, low confidence |
| Toolathlon-Verified | 60.1% | estimated ± 3.0 pp, high confidence |
| AA AutomationBench | 40.9% | estimated ± 9.9 pp, low confidence |
| GDP.pdf | 11.8% | estimated ± 4.2 pp, medium confidence |
| AA Terminal-Bench 4.0 | 0.0% | estimated ± 13.7 pp, low confidence |
| ARC-AGI-3 | 0.0% | estimated ± 8.6 pp, low confidence |
| CWE-bench | 30.1% | estimated ± 6.2 pp, low confidence |
| MKA | 32.8% | estimated ± 4.6 pp, medium confidence |
| SkillsBench official leaderboard (BenchFlow) | 20.1% | measured |
| BinaryAudit | 7.4% | estimated ± 12.0 pp, low confidence |
| OTelBench | 6.8% | estimated ± 6.6 pp, medium confidence |
| Vibe Code Bench | 0.0% | measured |
| Hermes Index | 17.8% | estimated ± 5.7 pp, low confidence |
| ParseBench-2 | 63.6% | measured |
| CharXiv | 73.2% | measured |
| CharXiv w/o tools | 75.5% | estimated ± 1.8 pp, low confidence |
| ExploitGym | 0.0% | estimated ± 8.3 pp, low confidence |
| AI2D_TEST | 87.8% | estimated ± 0.9 pp, medium confidence |
| CountBench | 93.2% | estimated ± 3.7 pp, low confidence |
| DynaMath | 82.7% | estimated ± 0.8 pp, medium confidence |
| ERQA | 61.9% | estimated ± 5.6 pp, medium confidence |
| HMMT Feb 2026 | 84.6% | estimated ± 3.1 pp, high confidence |
| HMMT Nov 2025 | 91.5% | estimated ± 1.5 pp, medium confidence |
| MathVision | 81.1% | estimated ± 3.3 pp, high confidence |
| MLVU (M-Avg) | 83.4% | estimated ± 1.5 pp, high confidence |
| MMAnswerBench | 80.9% | estimated ± 2.4 pp, medium confidence |
| MMMU-Pro | 75.2% | estimated ± 3.1 pp, high confidence |
| NL2Repo | 37.8% | estimated ± 4.7 pp, medium confidence |
| PostTrainBench v1 | 19.0% | estimated ± 3.1 pp, low confidence |
| RealWorldQA | 80.4% | estimated ± 1.5 pp, high confidence |
| ScreenSpot Pro | 58.6% | estimated ± 6.8 pp, medium confidence |
| Terminal-Bench 2.0 | 44.4% | estimated ± 5.8 pp, medium confidence |
| Toolathlon | 37.8% | estimated ± 8.0 pp, medium confidence |
| Video-MME (with subtitle) | 83.3% | estimated ± 2.9 pp, high confidence |
| VideoMMMU | 84.4% | estimated ± 1.5 pp, medium confidence |
| We-Math | 84.5% | estimated ± 7.9 pp, low confidence |
| AA LiveCodeBench | 84.2% | estimated ± 2.6 pp, high confidence |
| MedXpertQA (MM) | 72.3% | estimated ± 6.4 pp, low confidence |
| React Native Evals | 0.0% | estimated ± 3.0 pp, low confidence |
| SWE-Rebench | 0.0% | estimated ± 6.5 pp, low confidence |
| PostTrainBench | 7.7% | estimated ± 3.7 pp, medium confidence |
| Finance Agent v2 | 55.1% | estimated ± 2.4 pp, low confidence |
| LABBench2 | 81.0% | estimated ± 1.8 pp, low confidence |
| ScarfBench | 35.0% | estimated ± 12.3 pp, low confidence |
| Agents' Last Exam | 29.6% | estimated ± 9.7 pp, low confidence |
| OpenHarmony Bench v1.0 | 48.9% | estimated ± 3.2 pp, medium confidence |
| BabyVision | 26.4% | estimated ± 11.5 pp, low confidence |
| MMVU | 73.1% | estimated ± 3.6 pp, low confidence |
| LifeVal | 2.9% | estimated ± 5.0 pp, low confidence |
| MMMU-Pro w/ Python | 77.9% | estimated ± 2.7 pp, medium confidence |