Claude Opus 5.5 vs Gemini 4 Hydrogen
Claude Opus 5.5 leads by 1.9 capability points (≈95% interval −0.8 to +4.6): statistically tied · P(higher) ≈ 92%
Wins 43 of 71 shared benchmarks (24 lost, 4 tied)
Capability vs. relative cost
Claude Opus 5.5Gemini 4 HydrogenClosed weightsOpen weights
Benchmark wins43 – 24 · 71 shared, 4 tied
Comparing matching setups only (5 benchmarks excluded) · suites counted once
Area← Claude Opus 5.5Gemini 4 Hydrogen →
Data: Surge AI, Epoch AI, Arena and 25 othersatlas.village.dev
Margins
Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Gemini 4 Hydrogen minus Claude Opus 5.5.
| TxBench Oligo Discovery V1 (113 evaluations) | Science › Life sciences | 1.5 | 54.3 | +52.8 |
| TxBench Antibody Discovery V1 (100 evaluations) | Science › Life sciences | 8.3 | 57.7 | +49.3 |
| TxBench Preclinical Pharmacology V1 Clean Reruns | Science › Life sciences | 26.3 | 61.3 | +35.0 |
| Furniture Assembly Benchmark | Perception & media › Spatial & visual reasoning | 83.3 | 57.5 | −25.8 |
| DAYJOB: Healthcare (October 2026 rescoring) | Health & medicine | 48.3 | 24.0 | −24.3 |
| SWE-Marathon v1.2 | Software › Software engineering | 81.3 | 58.1 | −23.1 |
| Humanity's Sixth Sense | Perception & media › Spatial & visual reasoning | 44.6 | 63.4 | +18.8 |
| LMArena Code Arena WebDev | Software › Frontend & design | 1818 | 1703 | −15.4 |
| sudo L7 (October 2026 rescoring) | Software › Software engineering | 46.1 | 31.7 | −14.4 |
| DAYJOB: Law (November 2026 release) | Professional › Legal | 23.7 | 9.7 | −14.0 |
| Terminal-Bench-Science 0.1 | Science › Multidisciplinary | 61.9 | 49.5 | −12.4 |
| GDP.xlsx | Professional › General knowledge work | 30.3 | 40.9 | +10.6 |
| AA-Briefcase v1.1 Elo | Professional › General knowledge work | 1760 | 1569 | −10.5 |
| FrontierSWE V2 Best@5 | Software › Software engineering | 71.5 | 61.2 | −10.3 |
| LMArena Agent Arena Net Improvement (Steering Burden) | Tool & computer use › General agents | 18.6 | 13.5 | −10.0 |
| ComplexConstraints (November 2026 rescoring) | Language & writing › Instruction following | 64.4 | 73.7 | +9.3 |
| LMArena Agent Arena Bash Reliability | Tool & computer use › General agents | 14.0 | -4.5 | −9.3 |
| FrontierCode 2.0 - Pass Rate | Software › Software engineering | 51.8 | 43.2 | −8.6 |
| Bug Hunt Bench | Software › Software engineering | 39.7 | 31.1 | −8.6 |
| FrontierCode 2.0 - Score | Software › Software engineering | 47.3 | 38.9 | −8.5 |
| HANDBOOK.md Agents v1.1 | Tool & computer use › Customer service & operations | 52.3 | 44.6 | −7.7 |
| FrontierMath Tier 4 (v2) - Private | Math | 95.0 | 87.8 | −7.2 |
| Chess Puzzles (task version 2.0) | Games | 67.0 | 60.0 | −7.0 |
| FrontierSWE V2 Worst@5 | Software › Software engineering | 52.0 | 45.3 | −6.6 |
| LiveBench 2.0 (2026-11-06 release) | Composite indexes | 83.4 | 77.0 | −6.4 |
| GBENCH One-shot Coding GScore | Games | 73.0 | 66.8 | −6.2 |
| Terminal-Bench 4.0 (Artificial Analysis) | Software › Software engineering | 59.6 | 53.5 | −6.1 |
| Mystery Game Puzzles | Games | 71.0 | 65.0 | −6.0 |
| DAYJOB: Tax | Professional › Finance, accounting & tax | 9.2 | 3.6 | −5.6 |
| Professional Reasoning Benchmark - Legal | Professional › Legal | 47.6 | 53.1 | +5.5 |
| GDP.pdf (November 2026 rescoring) | Professional › General knowledge work | 48.8 | 54.2 | +5.4 |
| LMArena Agent Arena Praise vs Complaint | Tool & computer use › General agents | 38.6 | 29.3 | −4.7 |
| FORTRESS Average Risk Score (ARS) | Safety & alignment | 13.2 | 17.5 | −4.3 |
| Humanity's Last Exam Diamond | Knowledge & reasoning › General knowledge | 54.6 | 58.8 | +4.2 |
| DAYJOB: Finance (October 2026 rescoring) | Professional › Finance, accounting & tax | 42.5 | 38.5 | −4.0 |
| GDP.pdf (Artificial Analysis, All-pass) | Professional › General knowledge work | 28.8 | 32.8 | +4.0 |
| Ramp SWE-Bench v2 | Software › Software engineering | 67.8 | 71.6 | +3.8 |
| HiL-Bench (Human-in-Loop Benchmark) | Tool & computer use › General agents | 59.0 | 55.3 | −3.7 |
| CursorBench 4.0 | Software › Software engineering | 57.8 | 54.3 | −3.5 |
| SWE-sweep | Software › Software engineering | 10.1 | 13.5 | +3.4 |
| CursorBench 4.1 | Software › Software engineering | 58.1 | 61.4 | +3.3 |
| SimpleBench | Knowledge & reasoning › General knowledge | 88.4 | 91.7 | +3.3 |
| FrontierMath Tiers 1-3 (v2) - Private | Math | 91.2 | 88.1 | −3.2 |
| SWE-Marathon v1.1 | Software › Software engineering | 46.3 | 49.4 | +3.1 |
| LMArena Text Arena Elo | Language & writing › General assistance | 1507 | 1490 | −3.0 |
| Chartography (November 2026 rescoring) | Perception & media › Documents & charts | 89.6 | 87.2 | −2.4 |
| HLAB-Vals-1.1 | Professional › Legal | 9.2 | 11.3 | +2.1 |
| TaxCalcBench TY25 | Professional › Finance, accounting & tax | 42.0 | 44.0 | +2.0 |
| SpireClimb | Games | 5.8 | 7.7 | +1.8 |
| LMArena Agent Arena Confirmed Success | Tool & computer use › General agents | 19.2 | 15.7 | −1.8 |
| GradeBench | Professional › Education | 53.5 | 51.9 | −1.6 |
| Epoch Capabilities Index (ECI) | Composite indexes | 169 | 167 | −1.5 |
| SimpleQA Verified (Epoch anti-abstention) | Knowledge & reasoning › General knowledge | 72.2 | 70.9 | −1.3 |
| BigFinanceBench private holdout (final-answer accuracy) | Professional › Finance, accounting & tax | 48.2 | 49.4 | +1.2 |
| CWE-bench v2 Pass@1 | Software › Cybersecurity | 67.1 | 68.1 | +1.0 |
| DualEntry Accounting AI Benchmark | Professional › Finance, accounting & tax | 82.2 | 83.2 | +1.0 |
| Gray Swan IPI Attack Success (k=15) | Safety & alignment | 1.0 | 0.0 | +1.0 |
| EBR-bench v4 (card ban, single agent) | Games | 71.4 | 70.5 | −1.0 |
| CritPt v2 (agentic, avg@5) | Science › Physical sciences | 81.4 | 80.6 | −0.9 |
| ARC-AGI-2 | Knowledge & reasoning › Abstract reasoning | 93.3 | 92.5 | −0.8 |
| DAYJOB: Accounting | Professional › Finance, accounting & tax | 34.4 | 33.6 | −0.8 |
| micro1 PharmaBench | Professional › General knowledge work | 10.8 | 11.4 | +0.6 |
| Medical Long Context Reasoning (MLCR-AA, October 2026 methodology) | Health & medicine | 27.8 | 27.2 | −0.6 |
| LMArena Agent Arena Steering Burden | Tool & computer use › General agents | 18.0 | 17.4 | −0.3 |
| LMArena Agent Arena Tool Hallucination | Tool & computer use › Tool calling | 1.0 | 0.7 | −0.1 |
| BuseyBench SVG | Perception & media › Generation | 6.8 | 6.9 | +0.1 |
| Excel Modeling Benchmark v1.1 | Professional › Finance, accounting & tax | 75.8 | 75.8 | 0.0 |
| Agents' Last Exam v1.1 (ALE v1.1) | Professional › General knowledge work | 43.4 | 43.4 | 0.0 |
| ARC-AGI-1 Semi-Private | Knowledge & reasoning › Abstract reasoning | 98.5 | 98.5 | 0.0 |
| Dual-Arm Sim (task completion) | Tool & computer use › Robotics | 0.0 | 0.0 | 0.0 |
| OTIS Mock AIME 2024-2025 | Math | 100.0 | 100.0 | 0.0 |
Data: Surge AI, Epoch AI, Arena and 25 othersatlas.village.dev