Gemini 3.1 Flash-Lite ranks first on StructEval.
Google DeepMind Mar 2026 Closed weights Small
Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Global PIQA | 92% | #2 of 44 | Official leaderboard | |
| MMLU-Pro | 86.2% | #23 of 212 | Official leaderboard | |
| SEA-HELM | 75.3 | #13 of 71 | Official leaderboard | |
| Wer wird Millionär? LLM Benchmark | 114156.0 | #16 of 77 | Official leaderboard | |
| MMMLU | 88.9% | #15 of 71 | Lab model card | |
| LMArena (Elo) | 1,433 | #89 of 335 | Official leaderboard | |
| Judgemark v2.1 | 80.9 | #15 of 42 | Official leaderboard | |
| MultiChallenge | 60.6% | #16 of 42 | Official leaderboard | |
| Vectara Hallucination Leaderboard (2025 dataset) | 8.20% | #41 of 104 | Official leaderboard | |
| STALE | 22.4% | #4 of 10 | Official leaderboard | |
| Phare | 67.5% | #27 of 62 | Official leaderboard | |
| SimpleQA Verified | 43.3% | #10 of 22 | Lab model card | |
| FACTS Parametric | 48.4% | #21 of 45 | Official leaderboard | |
| StudyArena Text | 1,235 | #41 of 85 | Official leaderboard | |
| DesignArena ASCII Art | 1,162 | #38 of 77 | Official leaderboard | |
| Judgemark v4 | 58.8 | #20 of 40 | Official leaderboard | |
| FACTS Search | 66.8% | #18 of 31 | Official leaderboard | |
| MRCR v2 | 60.1% | #18 of 30 | Lab model card | |
| MultiNRC | 25% | #23 of 38 | Official leaderboard | |
| Humanity's Last Exam | 8.64% | #30 of 47 | Official leaderboard | |
| Longform Creative Writing (EQ-Bench) | 49.0 | #83 of 119 | Official leaderboard | |
| OpenAI MRCR v2 | 22.5% | #38 of 47 | Official leaderboard | |
| BullshitBench V2 | 11% | #112 of 129 | Official leaderboard | |
| BullshitBench | 9.09% | #99 of 107 | Official leaderboard | |
| Extended NYT Connections | 8.20% | #99 of 103 | Official leaderboard | |
| FACTS Benchmark Suite | 40.6% | #4 of 4 | Lab model card |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| GPQA Diamond | 81.8% | #95 of 229 | Epoch AI run | |
| OTIS Mock AIME 2024-2025 | 80% | #98 of 212 | Epoch AI run | |
| DTBench | 76.8% | #101 of 180 | Official leaderboard | |
| ForecastBench | 0.14 | #51 of 82 | Official leaderboard | |
| FrontierMath Tiers 1-3 v2 | 27.7% | #68 of 86 | Official leaderboard | |
| BixBench3 | 0.00 | #20 of 20 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| StructEval | 81.2% | #1 of 28 | Official leaderboard | |
| WeirdML | 52.2% | #46 of 112 | Official leaderboard | |
| ALE-Bench | 1,165 | #88 of 126 | Official leaderboard | |
| DesignArena Website | 1,076 | #121 of 151 | Official leaderboard | |
| DesignArena SVG | 1,043 | #83 of 103 | Official leaderboard | |
| DesignArena UI Component | 1,063 | #113 of 139 | Official leaderboard | |
| DesignArena Code | 1,063 | #118 of 143 | Official leaderboard | |
| DesignArena Game Dev | 1,020 | #122 of 143 | Official leaderboard | |
| DesignArena 3D Design | 1,031 | #117 of 136 | Official leaderboard | |
| Code Arena | 1,256 | #114 of 128 | Official leaderboard | |
| WebDev Arena | 1,254 | #94 of 105 | arena.ai, via Epoch AI | |
| DesignArena Data Visualization | 1,043 | #109 of 118 | Official leaderboard | |
| Code Migration | 4.61% | #73 of 78 | Official leaderboard | |
| Image-to-WebDev Arena | 1,338 | #54 of 55 | Official leaderboard | |
| Vibe Code Bench | 0% | #118 of 120 | Official leaderboard | |
| FrontierCode | 4.84% | #12 of 12 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| SnakeBench | 31 | #35 of 188 | Official leaderboard | |
| PinchBench 2.0 | 80.5% | #16 of 57 | Official leaderboard | |
| PACT | 1,509 | #16 of 26 | Official leaderboard | |
| LLM Chess | -133 | #83 of 123 | Official leaderboard | |
| Clocktower Radio | 1,405 | #18 of 23 | Official leaderboard | |
| Cooperate to Compete (C2C) | -0.42 | #4 of 5 | Official leaderboard | |
| MCP Atlas | 57.1% | #33 of 37 | Official leaderboard | |
| UniClawBench | 15.5% | #9 of 10 | Official leaderboard | |
| Deep Research Bench (FutureSearch) | 0.37 | #19 of 21 | Official leaderboard | |
| DeepResearch Bench | 37.3% | #20 of 22 | evals.futuresearch.ai, via Epoch AI | |
| Toolathlon | 14.2% | #47 of 50 | Official leaderboard | |
| SkillsBench | 20.1% | #22 of 22 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| MortgageTax | 68% | #19 of 87 | Official leaderboard | |
| MedCode | 47.6% | #31 of 96 | Official leaderboard | |
| LegalBench | 83.8% | #49 of 138 | Official leaderboard | |
| TaxEval v2 | 71.8% | #70 of 129 | Official leaderboard | |
| CorpFin v2 | 59.4% | #75 of 119 | Official leaderboard | |
| Vals Finance Agent | 46.1% | #33 of 49 | Official leaderboard | |
| CaseLaw v2 | 55% | #36 of 53 | Official leaderboard | |
| PolicyBench | 79.3% | #38 of 50 | Official leaderboard | |
| CADGenBench | 24.9% | #9 of 11 | Official leaderboard | |
| Harvey LAB (Legal Agent Benchmark) | 0% | #68 of 80 | Official leaderboard | |
| 3DCodeArena | 849 | #12 of 14 | Official leaderboard | |
| YC-Bench | $202,899 | #51 of 59 | Official leaderboard | |
| TRLawBench | 7.50 | #31 of 34 | Official leaderboard | |
| Vals Finance Agent v2 | 30% | #73 of 80 | Official leaderboard | |
| MedScribe | 63.9% | #90 of 97 | Official leaderboard | |
| Vals Index v1 | 40.8% | #32 of 34 | Official leaderboard | |
| Legal Research Bench | 3.37% | #89 of 93 | Official leaderboard | |
| Vals Index v2 | 15.5% | #63 of 65 | Official leaderboard | |
| Tax Agent Bench | 3.11% | #82 of 84 | Official leaderboard | |
| Excel Modeling Benchmark | 8.63% | #76 of 76 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Video-MMMU | 84.8% | #8 of 36 | Lab model card | |
| GeoBench | 83% | #11 of 48 | Official leaderboard | |
| MMMU-Pro | 76.8% | #33 of 90 | Lab model card | |
| VISTA | 46.9% | #20 of 54 | Official leaderboard | |
| Vision Arena | 1,235 | #57 of 136 | Official leaderboard | |
| 1H-VideoQA | 78.2% | #9 of 20 | Official leaderboard | |
| OVO-S-Bench | 50.8% | #7 of 15 | Official leaderboard | |
| CharXiv | 73.2% | #49 of 75 | Lab model card | |
| ParseBench | 56.7% | #29 of 44 | Official leaderboard | |
| ZeroBench | 5.00% | #54 of 76 | Official leaderboard | |
| vCX Benchmark | 82.6% | #32 of 42 | Official leaderboard | |
| MineBench | 944 | #71 of 75 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| nocot-bench (NCRI) | 86.5 | #41 of 185 | Official leaderboard | |
| BrowseSafe++ | 80.9% | #8 of 15 | Official leaderboard | |
| PCT (Polarized Contrastive Pairs) | 51.9% | #37 of 46 | Official leaderboard | |
| FORTRESS | 51% | #59 of 73 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| SAGE (Vals) | 49.5% | #20 of 81 | Official leaderboard | |
| Political Science LLM Benchmark | 68.1% | #11 of 42 | Official leaderboard | |
| ACCoRD | 64.1 | #78 of 152 | Official leaderboard | |
| LMCA | 35% | #78 of 147 | Official leaderboard | |
| Conceptual Reasoning Index | 46.1 | #83 of 151 | Official leaderboard | |
| TutorBench | 51.5% | #13 of 20 | Official leaderboard | |
| CAIS Vision Capabilities Index | 44.7 | #39 of 53 | Official leaderboard | |
| CAIS Risk Index | 61.0 | #53 of 67 | Official leaderboard | |
| CAIS Text Capabilities Index | 18.1 | #55 of 68 | Official leaderboard | |
| Vals Multimodal Index | 41.4% | #31 of 33 | Official leaderboard | |
| JevBench | 59.8% | #4 of 4 | Official leaderboard |
The Self-Reported Index fits one score to the 8 capability results Google DeepMind reported for this model, on the scale of the Epoch Capabilities Index. Labs choose which benchmarks to show and how to run them, so read it as the lab's claim, not an independent measurement. The 90% interval runs from 140.5 to 146.8. How we compute it.
Where the benchmark's own leaderboard, or Epoch AI's run, also has this model. Gaps mostly come from different harnesses, effort levels and task subsets. Every pair.
| Benchmark | Lab says | Board says | Gap |
|---|---|---|---|
| Humanity's Last Exam | 16% Humanity's Last Exam (full set, text + MM), No tools | 8.64% Leaderboard | +7.4 pts |
| GPQA Diamond | 86.9% GPQA Diamond (No tools) | 81.8% Epoch AI run, high | +5.1 pts |
| LMArena (Elo) | 1,432 Elo 1432 on the Arena.ai leaderboard, quoted in the Gemini 3.1 Flash-Lite launch blog | 1,433 Leaderboard | −1 |
Google DeepMind's models in color, against every other model on the Epoch Capabilities Index, by release date.
Source: Epoch Capabilities Index, Epoch AI (CC BY 4.0)
The benchmarks the lab chose to headline, with the scores it gave.
Every result in the longer documents the lab published with it. 5 of them aren't in the launch post.