Claude Sonnet 5.5 vs Gemini 4 Hydrogen
TryTop 2 now
Claude Sonnet 5.5 leads by 0.9 capability points (≈95% interval −1.8 to +3.5): statistically tied · P(higher) ≈ 74%
Gemini 4 Hydrogen wins 40 of 76 shared benchmarks (32 lost, 4 tied)
Capability vs. relative cost
Claude Sonnet 5.5Gemini 4 HydrogenClosed weightsOpen weights
Benchmark wins32 – 40 · 76 shared, 4 tied
Comparing matching setups only (2 benchmarks excluded) · suites counted once
Area← Claude Sonnet 5.5Gemini 4 Hydrogen →
Data: Artificial Analysis, Surge AI, Epoch AI and 24 othersatlas.village.dev
Margins
Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Gemini 4 Hydrogen minus Claude Sonnet 5.5.
| TaxCalcBench TY25 | Professional › Finance, accounting & tax | 18.0 | 44.0 | +26.0 |
| SimpleQA Verified (Epoch anti-abstention) | Knowledge & reasoning › General knowledge | 46.5 | 70.9 | +24.4 |
| TxBench-Antibody-Development V1 (60 evaluations) | Science › Life sciences | 36.7 | 60.6 | +23.9 |
| SWE-Marathon v1.2 | Software › Software engineering | 80.6 | 58.1 | −22.5 |
| sudo L7 (October 2026 rescoring) | Software › Software engineering | 50.6 | 31.7 | −18.9 |
| Bug Hunt Bench | Software › Software engineering | 48.9 | 31.1 | −17.8 |
| Furniture Assembly Benchmark | Perception & media › Spatial & visual reasoning | 75.0 | 57.5 | −17.5 |
| DAYJOB: Law (November 2026 release) | Professional › Legal | 26.8 | 9.7 | −17.1 |
| SimpleBench | Knowledge & reasoning › General knowledge | 75.9 | 91.7 | +15.8 |
| Medical Long Context Reasoning (MLCR-AA, October 2026 methodology) | Health & medicine | 41.7 | 27.2 | −14.4 |
| Artificial Analysis Intelligence Index v5.0.1 | Composite indexes | 54.5 | 41.2 | −13.3 |
| GDP.xlsx | Professional › General knowledge work | 29.1 | 40.9 | +11.8 |
| Agents' Last Exam v1.1 Score | Professional › General knowledge work | 60.4 | 71.6 | +11.2 |
| DAYJOB: Healthcare (October 2026 rescoring) | Health & medicine | 34.0 | 24.0 | −10.0 |
| AA-Briefcase v1.1 Elo | Professional › General knowledge work | 1739 | 1569 | −9.3 |
| LMArena Code Arena WebDev | Software › Frontend & design | 1773 | 1703 | −9.3 |
| FrontierSWE V2 | Software › Software engineering | 61.9 | 53.1 | −8.8 |
| ARC-AGI-2 | Knowledge & reasoning › Abstract reasoning | 83.9 | 92.5 | +8.6 |
| Ramp SWE-Bench v2 | Software › Software engineering | 63.2 | 71.6 | +8.4 |
| LMArena Agent Arena Bash Reliability | Tool & computer use › General agents | 11.9 | -4.5 | −8.2 |
| GradeBench | Professional › Education | 43.7 | 51.9 | +8.2 |
| BigFinanceBench private holdout (final-answer accuracy) | Professional › Finance, accounting & tax | 41.5 | 49.4 | +7.9 |
| AA-Omniscience v1.1 Index | Knowledge & reasoning › General knowledge | 30.5 | 46.3 | +7.9 |
| FrontierMath Tier 4 (v2) - Private | Math | 80.5 | 87.8 | +7.3 |
| DAYJOB: Tax | Professional › Finance, accounting & tax | 10.8 | 3.6 | −7.2 |
| GDP.pdf (Artificial Analysis, All-pass) | Professional › General knowledge work | 25.8 | 32.8 | +7.0 |
| CursorBench 4.1 | Software › Software engineering | 54.4 | 61.4 | +7.0 |
| CWE-bench v2 Pass@1 | Software › Cybersecurity | 61.4 | 68.1 | +6.7 |
| ComplexConstraints (November 2026 rescoring) | Language & writing › Instruction following | 67.2 | 73.7 | +6.5 |
| SpireClimb | Games | 1.3 | 7.7 | +6.3 |
| Humanity's Last Exam Diamond | Knowledge & reasoning › General knowledge | 53.3 | 58.8 | +5.5 |
| CritPt v2 (agentic, avg@5) | Science › Physical sciences | 75.1 | 80.6 | +5.4 |
| Chartography | Perception & media › Documents & charts | 61.6 | 66.6 | +5.0 |
| Vibe Code Bench v1.1 | Software › Frontend & design | 92.4 | 87.8 | −4.6 |
| Artificial Analysis Legal Index | Professional › Legal | 52.2 | 56.7 | +4.6 |
| LMArena Agent Arena Net Improvement (Steering Burden) | Tool & computer use › General agents | 15.6 | 13.5 | −4.0 |
| Terminal-Bench 4.0 (Artificial Analysis) | Software › Software engineering | 57.1 | 53.5 | −3.5 |
| Gray Swan IPI Attack Success (k=15) | Safety & alignment | 3.4 | 0.0 | +3.4 |
| FrontierCode 2.0 - Score | Software › Software engineering | 42.2 | 38.9 | −3.4 |
| BuseyBench SVG | Perception & media › Generation | 6.5 | 6.9 | +3.3 |
| FrontierCode 2.0 - Pass Rate | Software › Software engineering | 46.4 | 43.2 | −3.2 |
| HANDBOOK.md Agents v1.1 | Tool & computer use › Customer service & operations | 47.7 | 44.6 | −3.1 |
| TxBench Antibody Discovery V1 (100 evaluations) | Science › Life sciences | 54.7 | 57.7 | +3.0 |
| DualEntry Accounting AI Benchmark | Professional › Finance, accounting & tax | 80.2 | 83.2 | +3.0 |
| GDP.pdf (November 2026 rescoring) | Professional › General knowledge work | 51.2 | 54.2 | +3.0 |
| TxBench Preclinical Pharmacology V1 Clean Reruns | Science › Life sciences | 64.3 | 61.3 | −3.0 |
| Terminal-Bench-Science 0.1 | Science › Multidisciplinary | 52.4 | 49.5 | −2.9 |
| LMArena Text Arena Elo | Language & writing › General assistance | 1474 | 1490 | +2.8 |
| AutomationBench-AA 1.0.6 | Tool & computer use › Customer service & operations | 71.8 | 74.4 | +2.7 |
| TxBench Oligo Discovery V1 (113 evaluations) | Science › Life sciences | 51.9 | 54.3 | +2.4 |
| Artificial Analysis Finance & Accounting Index | Professional › Finance, accounting & tax | 55.8 | 58.1 | +2.3 |
| ARC-AGI-1 Semi-Private | Knowledge & reasoning › Abstract reasoning | 96.5 | 98.5 | +2.0 |
| DAYJOB: Finance (October 2026 rescoring) | Professional › Finance, accounting & tax | 40.5 | 38.5 | −2.0 |
| SWE-Marathon v1.1 | Software › Software engineering | 51.3 | 49.4 | −1.9 |
| Artificial Analysis Engineering Index | Science › Engineering | 58.5 | 60.1 | +1.6 |
| DAYJOB: Accounting | Professional › Finance, accounting & tax | 35.2 | 33.6 | −1.6 |
| LMArena Agent Arena Steering Burden | Tool & computer use › General agents | 14.8 | 17.4 | +1.3 |
| CursorBench 4.0 | Software › Software engineering | 55.5 | 54.3 | −1.2 |
| Chess Puzzles (task version 2.0) | Games | 59.0 | 60.0 | +1.0 |
| Artificial Analysis Economics Index | Composite indexes | 54.1 | 55.0 | +1.0 |
| LMArena Agent Arena Confirmed Success | Tool & computer use › General agents | 17.6 | 15.7 | −0.9 |
| Artificial Analysis Healthcare & Medical Index | Health & medicine | 54.2 | 53.3 | −0.9 |
| HLAB-Vals-1.1 | Professional › Legal | 10.4 | 11.3 | +0.8 |
| FORTRESS Average Risk Score (ARS) | Safety & alignment | 18.2 | 17.5 | +0.7 |
| LMArena Agent Arena Praise vs Complaint | Tool & computer use › General agents | 30.7 | 29.3 | −0.7 |
| FrontierMath Tiers 1-3 (v2) - Private | Math | 88.8 | 88.1 | −0.7 |
| GBENCH One-shot Coding GScore | Games | 66.4 | 66.8 | +0.4 |
| Artificial Analysis Strategy & Ops Index | Professional › General knowledge work | 58.1 | 57.7 | −0.3 |
| LiveBench 2.0 (2026-11-06 release) | Composite indexes | 76.8 | 77.0 | +0.2 |
| Epoch Capabilities Index (ECI) | Composite indexes | 167 | 167 | +0.2 |
| LMArena Agent Arena Tool Hallucination | Tool & computer use › Tool calling | 1.0 | 0.7 | −0.1 |
| Excel Modeling Benchmark v1.1 | Professional › Finance, accounting & tax | 75.9 | 75.8 | −0.1 |
| Dual-Arm Sim (task completion) | Tool & computer use › Robotics | 0.0 | 0.0 | 0.0 |
| micro1 PharmaBench | Professional › General knowledge work | 11.4 | 11.4 | 0.0 |
| Mystery Game Puzzles | Games | 65.0 | 65.0 | 0.0 |
| OTIS Mock AIME 2024-2025 | Math | 100.0 | 100.0 | 0.0 |
Data: Artificial Analysis, Surge AI, Epoch AI and 24 othersatlas.village.dev