Comparing 2 models
Claude Sonnet 5.5 vs Gemini 3.8 Flash
Benchmarks through
Verdict
Gemini 3.8 Flash places higher on 1 of the 1 tasks where both are ranked. Gemini 3.8 Flash costs 2.4x less per input token. Gemini 3.8 Flash accepts 1.0x more context.
- Choose Gemini 3.8 Flash for RAG & Long Context (#6).
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 | Claude Sonnet 5.5 | Gemini 3.8 Flash | Source |
|---|---|---|---|
| Design Arena Coding | 1345 Elo | 1292 Elo | source · 2026-11-15 |
| Design Arena UI | 1389 Elo | 1303 Elo | source · 2026-11-15 |
| Design Arena Web-app Agent | 1295 Elo | 1213 Elo | source · 2026-11-15 |
| Design Arena Website | 1309 Elo | 1291 Elo | source · 2026-11-15 |
| FACTS Grounding | 62.5% | 68.2% | source · 2026-11-12 |
| FrontierMath Tiers 1-3 (Epoch AI) | 88.8% ± 1.9 | 68.4% ± 2.8 | source · 2026-09-29 |
| GPQA Diamond (Epoch AI) | 95.6% ± 1.4 | 95.4% ± 1.4 | source · 2026-09-29 |
| LiveBench Agentic Coding | 56.3% | 54.2% | source · 2026-10-18 |
| LiveBench Coding | 91.4% | 72.5% | source · 2026-10-18 |
| LiveBench Data Analysis | 59.5% | 54.0% | source · 2026-10-18 |
| LiveBench Instruction Following | 56.8% | 81.4% | source · 2026-10-18 |
| LiveBench Language | 78.0% | 87.8% | source · 2026-10-18 |
| LiveBench Math | 96.1% | 91.6% | source · 2026-10-18 |
| LiveBench Reasoning | 91.6% | 89.3% | source · 2026-10-18 |
| LMArena Agent | 16 | 2.3 | source · 2026-11-11 |
| LMArena Agent Bash Recovery | 15 | -0.7 | source · 2026-10-08 |
| LMArena Agent Steerability | 14.5 | -1.3 | source · 2026-11-11 |
| LMArena Agent Task Outcome | 19.5 | 1.3 | source · 2026-11-11 |
| LMArena Agent Tool Hallucination | 0.9 | 0.4 | source · 2026-11-11 |
| LMArena Creative Writing | 1466 Elo | 1493 Elo | source · 2026-11-11 |
| LMArena Instruction Following | 1487 Elo | 1493 Elo | source · 2026-11-11 |
| LMArena Long Query | 1487 Elo | 1508 Elo | source · 2026-11-11 |
| LMArena Math | 1483 Elo | 1525 Elo | source · 2026-11-11 |
| LMArena Text | 1473 Elo | 1497 Elo | source · 2026-11-11 |
| LMArena Text Factuality | 1472 Elo | 1470 Elo | source · 2026-11-11 |
| LMArena Vision | 1263 Elo | 1293 Elo | source · 2026-11-09 |
| LMArena Vision Diagrams | 1308 Elo | 1342 Elo | source · 2026-11-09 |
| LMArena Vision OCR | 1281 Elo | 1315 Elo | source · 2026-11-09 |
| LMArena WebDev | 1772 Elo | 1583 Elo | source · 2026-11-11 |
| LMArena WebDev Image-to-Web | 1744 Elo | 1594 Elo | source · 2026-11-10 |
| OTIS Mock AIME (Epoch AI) | 100% ± 0 | 98.9% ± 0.9 | source · 2026-09-29 |
| SimpleQA Verified (Epoch AI) | 46.5% ± 1.6 | 69.7% ± 1.5 | source · 2026-09-29 |
| Terminal-Bench (4.1) | 63.9% | 18.2% | source · 2026-11-15 |
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.