Gemini 4 Hydrogen vs GPT-6.1 Terra
Gemini 4 Hydrogen leads by 0.5 capability points (≈95% interval −2.1 to +3.0): statistically tied · P(higher) ≈ 64%
Wins 45 of 61 shared benchmarks (15 lost, 1 tied)
Capability vs. relative cost
Gemini 4 HydrogenGPT-6.1 TerraClosed weightsOpen weights
Benchmark wins45 – 15 · 61 shared, 1 tied
Comparing matching setups only (31 benchmarks excluded) · suites counted once
Area← Gemini 4 HydrogenGPT-6.1 Terra →
Data: Artificial Analysis, Vals AI, Epoch AI and 12 othersatlas.village.dev
Margins
Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is GPT-6.1 Terra minus Gemini 4 Hydrogen.
| TxBench-Antibody-Development V1 (60 evaluations) | Science › Life sciences | 60.6 | 22.8 | −37.8 |
| Legal Research Bench | Professional › Legal | 53.8 | 26.0 | −27.9 |
| TxBench Oligo Discovery V1 (113 evaluations) | Science › Life sciences | 54.3 | 30.4 | −23.9 |
| Tax Agent Bench | Professional › Finance, accounting & tax | 78.4 | 55.0 | −23.4 |
| Furniture Assembly Benchmark | Perception & media › Spatial & visual reasoning | 57.5 | 79.2 | +21.7 |
| TaxCalcBench TY25 | Professional › Finance, accounting & tax | 44.0 | 24.0 | −20.0 |
| SimpleBench | Knowledge & reasoning › General knowledge | 91.7 | 72.2 | −19.5 |
| TxBench Antibody Discovery V1 (100 evaluations) | Science › Life sciences | 57.7 | 40.7 | −17.0 |
| Vals Legal Index | Professional › Legal | 73.6 | 58.8 | −14.9 |
| SpireClimb | Games | 7.7 | 22.3 | +14.7 |
| AutomationBench-AA 1.0.6 | Tool & computer use › Customer service & operations | 74.4 | 60.2 | −14.3 |
| SimpleQA Verified (Epoch anti-abstention) | Knowledge & reasoning › General knowledge | 70.9 | 57.3 | −13.6 |
| AA-Briefcase v1.1 Elo | Professional › General knowledge work | 1569 | 1334 | −12.9 |
| Vals Finance Index | Professional › Finance, accounting & tax | 66.1 | 53.3 | −12.8 |
| HiL-Bench (Human-in-Loop Benchmark) | Tool & computer use › General agents | 55.3 | 68.0 | +12.7 |
| Humanity's Last Exam Diamond | Knowledge & reasoning › General knowledge | 58.8 | 47.3 | −11.5 |
| DrugDiscoveryBench | Science › Life sciences | 69.1 | 58.1 | −11.0 |
| Code Migration v1.1 (Vals Index v2.2 subset) | Software › Software engineering | 82.7 | 72.2 | −10.5 |
| Artificial Analysis Legal Index | Professional › Legal | 56.7 | 46.4 | −10.4 |
| Ramp SWE-Bench v2 (Bash harness) | Software › Software engineering | 70.8 | 60.6 | −10.2 |
| Artificial Analysis Intelligence Index v5.0.1 | Composite indexes | 41.2 | 51.3 | +10.1 |
| GBENCH One-shot Coding GScore | Games | 66.8 | 76.4 | +9.6 |
| HLAB-Vals-1.1 | Professional › Legal | 11.3 | 1.7 | −9.6 |
| Finance Agent v2.1 | Professional › Finance, accounting & tax | 65.1 | 56.0 | −9.2 |
| Artificial Analysis Finance & Accounting Index | Professional › Finance, accounting & tax | 58.1 | 49.0 | −9.1 |
| Mystery Game Puzzles | Games | 65.0 | 74.0 | +9.0 |
| Artificial Analysis Healthcare & Medical Index | Health & medicine | 53.3 | 44.7 | −8.6 |
| LMArena Agent Arena Bash Reliability | Tool & computer use › General agents | -4.5 | 12.7 | +8.6 |
| AA-Omniscience v1.1 Index | Knowledge & reasoning › General knowledge | 46.3 | 29.6 | −8.3 |
| Vals Index v2.2 | Composite indexes | 68.5 | 60.5 | −8.0 |
| Artificial Analysis Strategy & Ops Index | Professional › General knowledge work | 57.7 | 50.0 | −7.7 |
| EBR-bench v4 (card ban, single agent) | Games | 70.5 | 78.1 | +7.6 |
| Terminal-Bench-Science 0.1 | Science › Multidisciplinary | 49.5 | 41.9 | −7.6 |
| Gray Swan IPI Attack Success (k=15) | Safety & alignment | 0.0 | 7.3 | −7.3 |
| Terminal-Bench 4.0 (Artificial Analysis) | Software › Software engineering | 53.5 | 46.5 | −7.1 |
| Artificial Analysis Economics Index | Composite indexes | 55.0 | 48.5 | −6.6 |
| MultiChallenge | Language & writing › Conversation | 80.6 | 74.2 | −6.4 |
| LMArena Agent Arena Net Improvement (Steering Burden) | Tool & computer use › General agents | 13.5 | 10.5 | −5.9 |
| LMArena Agent Arena Praise vs Complaint | Tool & computer use › General agents | 29.3 | 17.9 | −5.7 |
| TxBench Preclinical Pharmacology V1 Clean Reruns | Science › Life sciences | 61.3 | 55.7 | −5.7 |
| Artificial Analysis Engineering Index | Science › Engineering | 60.1 | 54.7 | −5.4 |
| DualEntry Accounting AI Benchmark | Professional › Finance, accounting & tax | 83.2 | 78.2 | −5.0 |
| FrontierMath Tier 4 (v2) - Private | Math | 87.8 | 92.7 | +4.9 |
| Terminal-Bench 4.0 | Software › Software engineering | 54.0 | 49.5 | −4.5 |
| GDP.pdf (Artificial Analysis, All-pass) | Professional › General knowledge work | 32.8 | 28.6 | −4.2 |
| BuseyBench SVG | Perception & media › Generation | 6.9 | 7.3 | +4.2 |
| Vibe Code Bench v1.1 | Software › Frontend & design | 87.8 | 84.0 | −3.8 |
| LMArena Agent Arena Confirmed Success | Tool & computer use › General agents | 15.7 | 8.5 | −3.6 |
| CritPt v2 (agentic, avg@5) | Science › Physical sciences | 80.6 | 77.1 | −3.4 |
| Chartography | Perception & media › Documents & charts | 66.6 | 63.2 | −3.4 |
| Medical Long Context Reasoning (MLCR-AA, October 2026 methodology) | Health & medicine | 27.2 | 30.6 | +3.3 |
| FrontierMath Tiers 1-3 (v2) - Private | Math | 88.1 | 90.9 | +2.8 |
| Excel Modeling Benchmark v1.1 | Professional › Finance, accounting & tax | 75.8 | 73.3 | −2.5 |
| GPQA Diamond | Science › Multidisciplinary | 96.8 | 94.3 | −2.5 |
| FrontierCode 2.0 (Artificial Analysis) | Software › Software engineering | 39.0 | 36.9 | −2.1 |
| LMArena Agent Arena Steering Burden | Tool & computer use › General agents | 17.4 | 13.2 | −2.1 |
| LiveBench 2.0 (2026-11-06 release) | Composite indexes | 77.0 | 78.6 | +1.6 |
| Epoch Capabilities Index (ECI) | Composite indexes | 167 | 168 | +0.7 |
| FORTRESS Average Risk Score (ARS) | Safety & alignment | 17.5 | 18.2 | −0.7 |
| LMArena Agent Arena Tool Hallucination | Tool & computer use › Tool calling | 0.7 | 1.2 | +0.2 |
| OTIS Mock AIME 2024-2025 | Math | 100.0 | 100.0 | 0.0 |
Data: Artificial Analysis, Vals AI, Epoch AI and 12 othersatlas.village.dev