Gemini 4 Carbon vs Qwen4-Max
Gemini 4 Carbon leads by 4.0 capability points (≈95% interval +1.4 to +6.7): a clear lead over Qwen4-Max · P(higher) ≈ 100%
Wins 40 of 50 shared benchmarks (9 lost, 1 tied)
Capability vs. relative cost
Gemini 4 CarbonQwen4-MaxClosed weightsOpen weights
Benchmark wins40 – 9 · 50 shared, 1 tied
Comparing matching setups only (4 benchmarks excluded) · suites counted once
Area← Gemini 4 CarbonQwen4-Max →
Data: Artificial Analysis, Surge AI, Vals AI and 4 othersatlas.village.dev
Margins
Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Qwen4-Max minus Gemini 4 Carbon.
| Terminal-Bench-Science 0.1 | Science › Multidisciplinary | 62.4 | 23.3 | −39.0 |
| Humanity's Last Exam Diamond | Knowledge & reasoning › General knowledge | 69.1 | 44.9 | −24.2 |
| Code Migration v1.1 (Vals Index v2.2 subset) | Software › Software engineering | 80.1 | 57.3 | −22.8 |
| LMArena Agent Arena Net Improvement (Steering Burden) | Tool & computer use › General agents | 16.1 | 7.3 | −17.2 |
| Artificial Analysis Intelligence Index v5.0.1 | Composite indexes | 60.0 | 42.9 | −17.1 |
| CritPt v2 (agentic, avg@5) | Science › Physical sciences | 82.3 | 66.0 | −16.3 |
| Artificial Analysis Engineering Index | Science › Engineering | 65.3 | 50.1 | −15.3 |
| GDP.xlsx | Professional › General knowledge work | 38.9 | 23.7 | −15.2 |
| sudo L7 (October 2026 rescoring) | Software › Software engineering | 30.6 | 15.6 | −15.0 |
| Terminal-Bench 4.0 | Software › Software engineering | 63.1 | 48.5 | −14.6 |
| Legal Research Bench | Professional › Legal | 55.8 | 41.3 | −14.4 |
| Terminal-Bench 4.0 (Artificial Analysis) | Software › Software engineering | 62.6 | 48.5 | −14.1 |
| FrontierCode 2.0 (Artificial Analysis) | Software › Software engineering | 40.6 | 26.5 | −14.1 |
| LMArena Agent Arena Praise vs Complaint | Tool & computer use › General agents | 39.4 | 13.0 | −13.2 |
| AA-Omniscience v1.1 Index | Knowledge & reasoning › General knowledge | 53.5 | 27.7 | −12.9 |
| GDP.pdf (November 2026 rescoring) | Professional › General knowledge work | 56.4 | 43.6 | −12.8 |
| Artificial Analysis Legal Index | Professional › Legal | 62.0 | 49.6 | −12.4 |
| ComplexConstraints (November 2026 rescoring) | Language & writing › Instruction following | 70.9 | 59.1 | −11.8 |
| EnterpriseBench: CoreCraft | Tool & computer use › Customer service & operations | 80.5 | 68.7 | −11.8 |
| KradleBench | Games | 22.0 | 11.0 | −11.0 |
| DAYJOB: Healthcare (October 2026 rescoring) | Health & medicine | 35.2 | 24.4 | −10.8 |
| HANDBOOK.md Agents v1.1 | Tool & computer use › Customer service & operations | 49.8 | 39.4 | −10.4 |
| LiveBench 2.0 (2026-11-06 release) | Composite indexes | 83.3 | 73.1 | −10.3 |
| Chartography (November 2026 rescoring) | Perception & media › Documents & charts | 85.5 | 75.3 | −10.2 |
| Artificial Analysis Healthcare & Medical Index | Health & medicine | 56.8 | 48.2 | −8.6 |
| Artificial Analysis Finance & Accounting Index | Professional › Finance, accounting & tax | 62.6 | 54.1 | −8.5 |
| Artificial Analysis Economics Index | Composite indexes | 59.6 | 51.3 | −8.3 |
| Tax Agent Bench | Professional › Finance, accounting & tax | 76.4 | 70.0 | −6.4 |
| Excel Modeling Benchmark v1.1 | Professional › Finance, accounting & tax | 76.5 | 70.8 | −5.6 |
| AutomationBench-AA 1.0.6 | Tool & computer use › Customer service & operations | 77.8 | 72.6 | −5.3 |
| Vals Legal Index | Professional › Legal | 75.9 | 70.7 | −5.2 |
| Vals Finance Index | Professional › Finance, accounting & tax | 67.0 | 62.0 | −5.1 |
| LMArena Agent Arena Steering Burden | Tool & computer use › General agents | 17.9 | 8.0 | −4.9 |
| LMArena Agent Arena Bash Reliability | Tool & computer use › General agents | -5.7 | 3.4 | +4.5 |
| LMArena Agent Arena Confirmed Success | Tool & computer use › General agents | 17.2 | 8.8 | −4.2 |
| AA-Briefcase v1.1 Elo | Professional › General knowledge work | 1611 | 1686 | +4.1 |
| Artificial Analysis Strategy & Ops Index | Professional › General knowledge work | 60.9 | 56.9 | −4.1 |
| DAYJOB: Tax | Professional › Finance, accounting & tax | 4.8 | 8.8 | +4.0 |
| HLAB-Vals-1.1 | Professional › Legal | 17.5 | 21.3 | +3.8 |
| DAYJOB: Law (November 2026 release) | Professional › Legal | 13.8 | 17.5 | +3.7 |
| GDP.pdf (Artificial Analysis, All-pass) | Professional › General knowledge work | 29.6 | 26.2 | −3.4 |
| GDPval-AA v2.1 | Professional › General knowledge work | 1733 | 1809 | +3.4 |
| DAYJOB: Accounting | Professional › Finance, accounting & tax | 25.2 | 22.4 | −2.8 |
| LMArena Code Arena WebDev | Software › Frontend & design | 1769 | 1750 | −2.5 |
| Vibe Code Bench v1.1 | Software › Frontend & design | 89.1 | 86.7 | −2.4 |
| DAYJOB: Finance (October 2026 rescoring) | Professional › Finance, accounting & tax | 23.8 | 25.3 | +1.5 |
| Finance Agent v2.1 | Professional › Finance, accounting & tax | 67.2 | 66.7 | −0.5 |
| LLM Debate Benchmark | Language & writing › Language & wordplay | 1676 | 1679 | +0.4 |
| LMArena Agent Arena Tool Hallucination | Tool & computer use › Tool calling | 0.8 | 1.2 | +0.2 |
| Medical Long Context Reasoning (MLCR-AA, October 2026 methodology) | Health & medicine | 19.4 | 19.4 | 0.0 |
Data: Artificial Analysis, Surge AI, Vals AI and 4 othersatlas.village.dev