Comparing 2 models
Claude Sonnet 5.5 vs Gemini 4 Carbon
Benchmarks through
Verdict
Gemini 4 Carbon places higher on 11 of the 14 tasks where both are ranked. Gemini 4 Carbon accepts 1.0x more context.
- Choose Claude Sonnet 5.5 for Coding (#3), Frontend & UI (#3), Tool & Function Calling (#6).
- Choose Gemini 4 Carbon for RAG & Long Context (#1), Math & Reasoning (#2), Competition Math (#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 | Claude Sonnet 5.5 | Gemini 4 Carbon | Source |
|---|---|---|---|
| Design Arena Coding | 1345 Elo | 1348 Elo | source · 2026-11-15 |
| Design Arena UI | 1389 Elo | 1347 Elo | source · 2026-11-15 |
| Design Arena Website | 1309 Elo | 1319 Elo | source · 2026-11-15 |
| FACTS Grounding | 62.5% | 77.2% | source · 2026-11-12 |
| FrontierMath Tiers 1-3 (Epoch AI) | 88.8% ± 1.9 | 90.5% ± 1.7 | source · 2026-09-29 |
| GPQA Diamond (Epoch AI) | 95.6% ± 1.4 | 96.5% ± 1.2 | source · 2026-09-29 |
| LMArena Agent | 16 | 17 | source · 2026-11-11 |
| LMArena Agent Steerability | 14.5 | 19.1 | source · 2026-11-11 |
| LMArena Agent Task Outcome | 19.5 | 18.4 | source · 2026-11-11 |
| LMArena Agent Tool Hallucination | 0.9 | 0.5 | source · 2026-11-11 |
| LMArena Creative Writing | 1466 Elo | 1533 Elo | source · 2026-11-11 |
| LMArena Instruction Following | 1487 Elo | 1539 Elo | source · 2026-11-11 |
| LMArena Long Query | 1487 Elo | 1549 Elo | source · 2026-11-11 |
| LMArena Text | 1473 Elo | 1526 Elo | source · 2026-11-11 |
| LMArena Text Factuality | 1472 Elo | 1521 Elo | source · 2026-11-11 |
| LMArena WebDev | 1772 Elo | 1768 Elo | source · 2026-11-11 |
| OTIS Mock AIME (Epoch AI) | 100% ± 0 | 100% ± 0 | source · 2026-09-29 |
| SimpleQA Verified (Epoch AI) | 46.5% ± 1.6 | 77.7% ± 1.3 | source · 2026-09-29 |
| Terminal-Bench (4.1) | 63.9% | 64.5% | 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.