Claude Haiku 5.5 vs Gemini 4 Hydrogen
TryTop 2 now
Gemini 4 Hydrogen leads by 4.5 capability points (≈95% interval +2.1 to +6.9): a clear lead over Claude Haiku 5.5 · P(higher) ≈ 100%
Wins 58 of 64 shared benchmarks (6 lost)
Capability vs. relative cost
Claude Haiku 5.5Gemini 4 HydrogenClosed weightsOpen weights
Benchmark wins6 – 58 · 64 shared
Comparing matching setups only (2 benchmarks excluded) · suites counted once
Area← Claude Haiku 5.5Gemini 4 Hydrogen →
Data: Artificial Analysis, Vals AI, Epoch AI and 16 othersatlas.village.dev
Margins
Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Gemini 4 Hydrogen minus Claude Haiku 5.5.
| SimpleQA Verified (Epoch anti-abstention) | Knowledge & reasoning › General knowledge | 23.8 | 70.9 | +47.1 |
| FrontierMath Tier 4 (v2) - Private | Math | 46.3 | 87.8 | +41.5 |
| TaxCalcBench TY25 | Professional › Finance, accounting & tax | 4.0 | 44.0 | +40.0 |
| AutomationBench 1.0.6 (Zapier) | Tool & computer use › Customer service & operations | 12.0 | 51.1 | +39.1 |
| AutomationBench-AA 1.0.6 | Tool & computer use › Customer service & operations | 36.0 | 74.4 | +38.5 |
| Mystery Game Puzzles | Games | 30.0 | 65.0 | +35.0 |
| SimpleBench | Knowledge & reasoning › General knowledge | 59.2 | 91.7 | +32.5 |
| Code Migration v1.1 (Vals Index v2.2 subset) | Software › Software engineering | 50.2 | 82.7 | +32.5 |
| Terminal-Bench-Science 0.1 | Science › Multidisciplinary | 18.1 | 49.5 | +31.4 |
| ARC-AGI-2 | Knowledge & reasoning › Abstract reasoning | 61.3 | 92.5 | +31.2 |
| Humanity's Last Exam Diamond | Knowledge & reasoning › General knowledge | 29.6 | 58.8 | +29.2 |
| LMArena Agent Arena Net Improvement (Steering Burden) | Tool & computer use › General agents | -0.6 | 13.5 | +27.5 |
| EBR-bench v4 (card ban, single agent) | Games | 43.8 | 70.5 | +26.7 |
| Terminal-Bench 4.0 (Artificial Analysis) | Software › Software engineering | 29.3 | 53.5 | +24.2 |
| BuseyBench SVG | Perception & media › Generation | 4.6 | 6.9 | +23.0 |
| Artificial Analysis Strategy & Ops Index | Professional › General knowledge work | 39.1 | 57.7 | +18.6 |
| Artificial Analysis Legal Index | Professional › Legal | 38.7 | 56.7 | +18.1 |
| Terminal-Bench 4.0 | Software › Software engineering | 36.4 | 54.0 | +17.7 |
| Artificial Analysis Engineering Index | Science › Engineering | 42.5 | 60.1 | +17.6 |
| LMArena Agent Arena Praise vs Complaint | Tool & computer use › General agents | -5.5 | 29.3 | +17.4 |
| AA-Omniscience v1.1 Index | Knowledge & reasoning › General knowledge | 12.3 | 46.3 | +17.0 |
| Tax Agent Bench | Professional › Finance, accounting & tax | 62.2 | 78.4 | +16.2 |
| Artificial Analysis Healthcare & Medical Index | Health & medicine | 37.5 | 53.3 | +15.8 |
| Artificial Analysis Finance & Accounting Index | Professional › Finance, accounting & tax | 43.6 | 58.1 | +14.5 |
| CritPt v2 (agentic, avg@5) | Science › Physical sciences | 66.3 | 80.6 | +14.3 |
| LMArena Code Arena WebDev | Software › Frontend & design | 1596 | 1703 | +14.3 |
| Ramp SWE-Bench v2 | Software › Software engineering | 57.4 | 71.6 | +14.2 |
| CursorBench 4.1 | Software › Software engineering | 48.4 | 61.4 | +13.0 |
| FrontierMath Tiers 1-3 (v2) - Private | Math | 75.1 | 88.1 | +13.0 |
| MultiChallenge | Language & writing › Conversation | 67.7 | 80.6 | +12.9 |
| Artificial Analysis Economics Index | Composite indexes | 42.7 | 55.0 | +12.4 |
| GDP.pdf (Artificial Analysis, All-pass) | Professional › General knowledge work | 20.8 | 32.8 | +12.0 |
| LMArena Text Arena Elo | Language & writing › General assistance | 1421 | 1490 | +11.9 |
| GBENCH One-shot Coding GScore | Games | 55.4 | 66.8 | +11.5 |
| FrontierCode 2.0 - Pass Rate | Software › Software engineering | 31.9 | 43.2 | +11.3 |
| Bug Hunt Bench | Software › Software engineering | 20.5 | 31.1 | +10.6 |
| Excel Modeling Benchmark v1.1 | Professional › Finance, accounting & tax | 65.5 | 75.8 | +10.3 |
| Legal Research Bench | Professional › Legal | 43.8 | 53.8 | +10.1 |
| SWE-Marathon v1.2 | Software › Software engineering | 48.1 | 58.1 | +10.0 |
| Furniture Assembly Benchmark | Perception & media › Spatial & visual reasoning | 47.5 | 57.5 | +10.0 |
| FrontierCode 2.0 - Score | Software › Software engineering | 28.9 | 38.9 | +9.9 |
| Vals Index v2.2 | Composite indexes | 58.7 | 68.5 | +9.8 |
| ARC-AGI-1 Semi-Private | Knowledge & reasoning › Abstract reasoning | 89.0 | 98.5 | +9.5 |
| FrontierSWE V2 | Software › Software engineering | 43.8 | 53.1 | +9.3 |
| Vals Finance Index | Professional › Finance, accounting & tax | 57.4 | 66.1 | +8.7 |
| LMArena Agent Arena Steering Burden | Tool & computer use › General agents | 0.8 | 17.4 | +8.3 |
| LiveBench 2.0 (2026-11-06 release) | Composite indexes | 68.9 | 77.0 | +8.1 |
| FORTRESS Average Risk Score (ARS) | Safety & alignment | 9.5 | 17.5 | −8.0 |
| Medical Long Context Reasoning (MLCR-AA, October 2026 methodology) | Health & medicine | 35.0 | 27.2 | −7.8 |
| LMArena Agent Arena Confirmed Success | Tool & computer use › General agents | 0.2 | 15.7 | +7.7 |
| HLAB-Vals-1.1 | Professional › Legal | 4.2 | 11.3 | +7.1 |
| DualEntry Accounting AI Benchmark | Professional › Finance, accounting & tax | 76.2 | 83.2 | +7.0 |
| CursorBench 4.0 | Software › Software engineering | 48.4 | 54.3 | +5.9 |
| LMArena Agent Arena Bash Reliability | Tool & computer use › General agents | 6.6 | -4.5 | −5.5 |
| Epoch Capabilities Index (ECI) | Composite indexes | 160 | 167 | +5.2 |
| Chess Puzzles (task version 2.0) | Games | 55.0 | 60.0 | +5.0 |
| Artificial Analysis Intelligence Index v5.0.1 | Composite indexes | 36.5 | 41.2 | +4.7 |
| Vals Legal Index | Professional › Legal | 70.3 | 73.6 | +3.3 |
| Vibe Code Bench v1.1 | Software › Frontend & design | 90.4 | 87.8 | −2.7 |
| AA-Briefcase v1.1 Elo | Professional › General knowledge work | 1523 | 1569 | +2.5 |
| LMArena Agent Arena Tool Hallucination | Tool & computer use › Tool calling | -1.5 | 0.7 | +1.1 |
| OTIS Mock AIME 2024-2025 | Math | 98.9 | 100.0 | +1.1 |
| SWE-Marathon v1.1 | Software › Software engineering | 50.0 | 49.4 | −0.6 |
| Finance Agent v2.1 | Professional › Finance, accounting & tax | 65.4 | 65.1 | −0.3 |
Data: Artificial Analysis, Vals AI, Epoch AI and 16 othersatlas.village.dev