Comparing 2 models
Gemini 3.1 Pro Preview vs Gemini 3.7 Flash
Benchmarks through
Verdict
Gemini 3.7 Flash places higher on 9 of the 9 tasks where both are ranked. Gemini 3.7 Flash costs 2.7x less per input token.
- Choose Gemini 3.7 Flash for Chat & Roleplay (#2), Translation (#2), Literary Translation (#2).
Built from the published task rankings below; ranks come from each task's own weighted benchmarks, so a win on one task says nothing about another.
Measured head to head
Benchmarks both models were run on
| Evaluation | Gemini 3.1 Pro Preview | Gemini 3.7 Flash | Source |
|---|---|---|---|
| Design Arena Coding | 1220 Elo | 1298 Elo | source · 2026-11-15 |
| Design Arena Full-stack Agent | 1041 Elo | 1183 Elo | source · 2026-11-15 |
| Design Arena UI | 1215 Elo | 1287 Elo | source · 2026-11-15 |
| Design Arena Web-app Agent | 1103 Elo | 1202 Elo | source · 2026-11-15 |
| Design Arena Website | 1225 Elo | 1295 Elo | source · 2026-11-15 |
| FrontierMath Tiers 1-3 (Epoch AI) | 59.6% ± 2.9 | 71.6% ± 2.7 | source · 2026-06-11 |
| GPQA Diamond (Epoch AI) | 94.4% ± 1.6 | 94.8% ± 1.3 | source · 2026-08-06 |
| GPQA Diamond (Epoch AI) | 94.1% ± 1.7 | 94.8% ± 1.3 | source · 2026-02-20 |
| LiveBench Agentic Coding | 44.1% | 58.3% | source · 2026-10-18 |
| LiveBench Coding | 76.5% | 78.9% | source · 2026-10-18 |
| LiveBench Data Analysis | 78.5% | 68.0% | source · 2026-10-18 |
| LiveBench Instruction Following | 79.1% | 79.9% | source · 2026-10-18 |
| LiveBench Language | 85.4% | 85.5% | source · 2026-10-18 |
| LiveBench Math | 91.0% | 93.5% | source · 2026-10-18 |
| LiveBench Reasoning | 84% | 87.8% | source · 2026-10-18 |
| LMArena Agent | -4.5 | -2.2 | source · 2026-11-11 |
| LMArena Agent Bash Recovery | -23.3 | -5.4 | source · 2026-10-08 |
| LMArena Agent Steerability | -8.2 | -7.3 | source · 2026-11-11 |
| LMArena Agent Task Outcome | -11.3 | -3.5 | source · 2026-11-11 |
| LMArena Agent Tool Hallucination | -0.4 | 0.7 | source · 2026-11-11 |
| LMArena Creative Writing | 1483 Elo | 1485 Elo | source · 2026-11-11 |
| LMArena Instruction Following | 1465 Elo | 1485 Elo | source · 2026-11-11 |
| LMArena Long Query | 1482 Elo | 1494 Elo | source · 2026-11-11 |
| LMArena Math | 1485 Elo | 1506 Elo | source · 2026-11-11 |
| LMArena Text | 1487 Elo | 1489 Elo | source · 2026-11-11 |
| LMArena Text Factuality | 1472 Elo | 1476 Elo | source · 2026-11-11 |
| LMArena Vision | 1280 Elo | 1294 Elo | source · 2026-11-09 |
| LMArena Vision Diagrams | 1309 Elo | 1320 Elo | source · 2026-11-09 |
| LMArena Vision OCR | 1295 Elo | 1308 Elo | source · 2026-11-09 |
| LMArena WebDev | 1446 Elo | 1588 Elo | source · 2026-11-11 |
| LMArena WebDev Image-to-Web | 1480 Elo | 1600 Elo | source · 2026-11-10 |
| OTIS Mock AIME (Epoch AI) | 95.6% ± 3.1 | 97.2% ± 1.9 | source · 2026-08-06 |
| OTIS Mock AIME (Epoch AI) | 95.6% ± 3.1 | 97.2% ± 1.9 | source · 2026-02-20 |
| SimpleQA Verified (Epoch AI) | 73.5% ± 1.4 | 69.2% ± 1.5 | source · 2026-08-10 |
Values are as published by each evaluator, in its own units; rows from different evaluators are not on a common scale.
task evidence
How they place when the question changes
Lower ranks are better. “Best” marks the strongest option in this shortlist for that task.