Gemini 4 Argon vs Muse Ember
TryTop 2 now
Gemini 4 Argon leads by 3.5 capability points (≈95% interval +1.0 to +5.9): a clear lead over Muse Ember · P(higher) ≈ 100%
Wins 44 of 53 shared benchmarks (9 lost)
Capability vs. relative cost
Gemini 4 ArgonMuse EmberClosed weightsOpen weights
Benchmark wins44 – 9 · 53 shared
Comparing matching setups only (3 benchmarks excluded) · suites counted once
Area← Gemini 4 ArgonMuse Ember →
Data: Artificial Analysis, Vals AI, Arena and 10 othersatlas.village.dev
Margins
Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Muse Ember minus Gemini 4 Argon.
| Humanity's Last Exam Diamond | Knowledge & reasoning › General knowledge | 53.2 | 22.0 | −31.2 |
| Terminal-Bench-Science 0.1 | Science › Multidisciplinary | 48.6 | 20.0 | −28.6 |
| Code Migration v1.1 (Vals Index v2.2 subset) | Software › Software engineering | 74.4 | 47.1 | −27.3 |
| DAXBench v2 | Software › Data & ML | 88.3 | 62.8 | −25.5 |
| GDP.xlsx | Professional › General knowledge work | 38.3 | 15.4 | −22.9 |
| BuseyBench SVG | Perception & media › Generation | 7.6 | 5.4 | −21.6 |
| RiemannBench (November 2026 rescoring) | Math | 74.4 | 53.6 | −20.8 |
| AA-Omniscience v1.1 Index | Knowledge & reasoning › General knowledge | 46.1 | 7.7 | −19.2 |
| SimpleBench | Knowledge & reasoning › General knowledge | 91.7 | 72.7 | −19.0 |
| LMArena Agent Arena Net Improvement (Steering Burden) | Tool & computer use › General agents | 10.3 | 1.0 | −18.2 |
| Terminal-Bench 4.0 | Software › Software engineering | 57.6 | 39.4 | −18.2 |
| Artificial Analysis Intelligence Index v5.0.1 | Composite indexes | 50.4 | 36.0 | −14.4 |
| Terminal-Bench 4.0 (Artificial Analysis) | Software › Software engineering | 55.1 | 40.9 | −14.1 |
| SWE Atlas v1.1 - Overall | Software › Software engineering | 70.1 | 56.0 | −14.1 |
| GBENCH One-shot Coding GScore | Games | 69.7 | 56.4 | −13.3 |
| LMArena Agent Arena Praise vs Complaint | Tool & computer use › General agents | 23.8 | -2.7 | −13.2 |
| DrugDiscoveryBench | Science › Life sciences | 65.4 | 52.4 | −13.0 |
| HiL-Bench (Human-in-Loop Benchmark) | Tool & computer use › General agents | 62.0 | 74.0 | +12.0 |
| APEX-Agents 1.1 | Professional › General knowledge work | 87.4 | 75.7 | −11.7 |
| ComplexConstraints (November 2026 rescoring) | Language & writing › Instruction following | 70.3 | 58.8 | −11.5 |
| CritPt v2 (agentic, avg@5) | Science › Physical sciences | 72.9 | 62.3 | −10.6 |
| ParseBench-2 | Perception & media › Documents & charts | 84.2 | 74.5 | −9.7 |
| LMArena Code Arena WebDev | Software › Frontend & design | 1681 | 1608 | −9.6 |
| AutomationBench-AA 1.0.6 | Tool & computer use › Customer service & operations | 77.5 | 68.1 | −9.4 |
| Tax Agent Bench | Professional › Finance, accounting & tax | 79.0 | 71.4 | −7.7 |
| LMArena Agent Arena Confirmed Success | Tool & computer use › General agents | 12.6 | -2.0 | −7.3 |
| Artificial Analysis Coding Agent Index v2.0 | Software › Software engineering | 53.2 | 45.9 | −7.3 |
| HANDBOOK.md Agents v1.1 | Tool & computer use › Customer service & operations | 36.6 | 29.8 | −6.8 |
| DAYJOB: Healthcare (October 2026 rescoring) | Health & medicine | 23.6 | 17.2 | −6.4 |
| DAYJOB: Finance (October 2026 rescoring) | Professional › Finance, accounting & tax | 30.5 | 24.2 | −6.3 |
| LMArena Agent Arena Bash Reliability | Tool & computer use › General agents | -6.0 | 6.6 | +6.3 |
| Professional Reasoning Benchmark - Legal | Professional › Legal | 49.4 | 55.6 | +6.2 |
| Professional Reasoning Benchmark - Finance | Professional › Finance, accounting & tax | 49.9 | 56.0 | +6.1 |
| Artificial Analysis Coding Agent Index v2.1 | Software › Software engineering | 45.5 | 39.7 | −5.8 |
| Legal Research Bench | Professional › Legal | 54.8 | 49.0 | −5.8 |
| Vals Finance Index | Professional › Finance, accounting & tax | 66.0 | 60.9 | −5.1 |
| DeepFinance | Professional › Finance, accounting & tax | 41.4 | 36.5 | −4.9 |
| τ³-Banking (Mercor run, bm25_grep retrieval) | Tool & computer use › Customer service & operations | 58.1 | 53.3 | −4.8 |
| Harvey LAB (Mercor run, undisclosed ~213-task cohort) | Professional › Legal | 27.4 | 22.7 | −4.7 |
| Artificial Analysis Progress Index v1 | Composite indexes | 342 | 329 | −4.6 |
| Medical Long Context Reasoning (MLCR-AA, October 2026 methodology) | Health & medicine | 13.9 | 18.3 | +4.4 |
| GDP.pdf (November 2026 rescoring) | Professional › General knowledge work | 53.6 | 49.2 | −4.4 |
| Finance Agent v2.1 | Professional › Finance, accounting & tax | 67.7 | 63.8 | −3.8 |
| LMArena Agent Arena Steering Burden | Tool & computer use › General agents | 12.2 | 4.8 | −3.7 |
| APEX-SWE | Software › Software engineering | 49.2 | 52.9 | +3.7 |
| GDP.pdf (Artificial Analysis, All-pass) | Professional › General knowledge work | 25.6 | 22.8 | −2.8 |
| Excel Modeling Benchmark v1.1 | Professional › Finance, accounting & tax | 74.8 | 72.2 | −2.7 |
| AA-Briefcase v1.1 Elo | Professional › General knowledge work | 1485 | 1526 | +2.3 |
| HLAB-Vals-1.1 | Professional › Legal | 27.1 | 25.0 | −2.1 |
| FrontierCode 2.0 (Artificial Analysis) | Software › Software engineering | 25.7 | 27.2 | +1.5 |
| Vals Legal Index | Professional › Legal | 76.8 | 75.7 | −1.1 |
| Ramp SWE-Bench v2 (Bash harness) | Software › Software engineering | 54.8 | 54.4 | −0.4 |
| LMArena Agent Arena Tool Hallucination | Tool & computer use › Tool calling | 0.8 | 0.9 | +0.1 |
Data: Artificial Analysis, Vals AI, Arena and 10 othersatlas.village.dev