Muse Ember vs Muse Spark 1.3
Muse Ember leads by 1.1 capability points (≈95% interval −1.1 to +3.4): statistically tied · P(higher) ≈ 84%
Muse Spark 1.3 wins 30 of 51 shared benchmarks (20 lost, 1 tied)
Capability vs. relative cost
Muse EmberMuse Spark 1.3Closed weightsOpen weights
Benchmark wins20 – 30 · 51 shared, 1 tied
Comparing matching setups only · suites counted once
Area← Muse EmberMuse Spark 1.3 →
Data: Artificial Analysis, Surge AI, Arena and 11 othersatlas.village.dev
Margins
Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Muse Spark 1.3 minus Muse Ember.
| DAXBench v2 | Software › Data & ML | 62.8 | 85.1 | +22.3 |
| Terminal-Bench 4.0 | Software › Software engineering | 39.4 | 24.7 | −14.6 |
| APEX-Agents 1.1 | Professional › General knowledge work | 75.7 | 61.3 | −14.4 |
| APEX-SWE | Software › Software engineering | 52.9 | 39.1 | −13.8 |
| AutomationBench-AA 1.0.6 | Tool & computer use › Customer service & operations | 68.1 | 57.9 | −10.2 |
| DrugDiscoveryBench | Science › Life sciences | 52.4 | 62.2 | +9.8 |
| SimpleBench | Knowledge & reasoning › General knowledge | 72.7 | 81.8 | +9.1 |
| Terminal-Bench-Science 0.1 | Science › Multidisciplinary | 20.0 | 11.0 | −9.0 |
| Humanity's Last Exam Diamond | Knowledge & reasoning › General knowledge | 25.1 | 33.7 | +8.6 |
| AA-Omniscience v1.1 Index | Knowledge & reasoning › General knowledge | 7.7 | 23.8 | +8.0 |
| Harvey LAB (Mercor run, undisclosed ~213-task cohort) | Professional › Legal | 22.7 | 15.0 | −7.7 |
| Terminal-Bench 4.0 (Artificial Analysis) | Software › Software engineering | 40.9 | 33.3 | −7.6 |
| FrontierCUA v1 pass@1 | Tool & computer use › Computer & browser use | 7.5 | 0.0 | −7.5 |
| GDP.xlsx | Professional › General knowledge work | 15.4 | 22.6 | +7.2 |
| HiL-Bench (Human-in-Loop Benchmark) | Tool & computer use › General agents | 74.0 | 66.9 | −7.1 |
| Legal Research Bench | Professional › Legal | 49.0 | 55.3 | +6.2 |
| LMArena Code Arena WebDev | Software › Frontend & design | 1608 | 1654 | +6.1 |
| ComplexConstraints (November 2026 rescoring) | Language & writing › Instruction following | 58.8 | 64.8 | +6.0 |
| Professional Reasoning Benchmark - Legal | Professional › Legal | 55.6 | 61.6 | +6.0 |
| CritPt v2 (agentic, avg@5) | Science › Physical sciences | 62.3 | 68.0 | +5.7 |
| Chartography (November 2026 rescoring) | Perception & media › Documents & charts | 79.3 | 84.6 | +5.3 |
| DAYJOB: Healthcare (October 2026 rescoring) | Health & medicine | 17.2 | 22.0 | +4.8 |
| HANDBOOK.md Agents v1.1 | Tool & computer use › Customer service & operations | 29.8 | 33.9 | +4.1 |
| FrontierCode 2.0 (Artificial Analysis) | Software › Software engineering | 27.2 | 23.1 | −4.1 |
| KradleBench | Games | 25.7 | 21.7 | −4.0 |
| TaxCalcBench TY25 | Professional › Finance, accounting & tax | 2.0 | 6.0 | +4.0 |
| GDP.pdf (Artificial Analysis, All-pass) | Professional › General knowledge work | 22.8 | 26.6 | +3.8 |
| GDP.pdf (November 2026 rescoring) | Professional › General knowledge work | 49.2 | 53.0 | +3.8 |
| Professional Reasoning Benchmark - Finance | Professional › Finance, accounting & tax | 56.0 | 59.5 | +3.6 |
| AA-Briefcase v1.1 Elo | Professional › General knowledge work | 1520 | 1581 | +3.3 |
| BuseyBench SVG | Perception & media › Generation | 5.4 | 5.7 | +3.1 |
| Artificial Analysis Coding Agent Index v2.1 | Software › Software engineering | 39.7 | 36.8 | −2.9 |
| DAYJOB: Finance (October 2026 rescoring) | Professional › Finance, accounting & tax | 24.2 | 27.0 | +2.8 |
| Medical Long Context Reasoning (MLCR-AA, October 2026 methodology) | Health & medicine | 18.3 | 15.6 | −2.8 |
| LMArena Agent Arena Bash Reliability | Tool & computer use › General agents | 6.6 | 1.2 | −2.7 |
| GBENCH One-shot Coding GScore | Games | 56.4 | 53.7 | −2.7 |
| APEX-1 (Mercor, mean score) | Professional › General knowledge work | 70.4 | 73.0 | +2.6 |
| LMArena Agent Arena Confirmed Success | Tool & computer use › General agents | -2.0 | 2.5 | +2.2 |
| Vibe Code Bench v1.1 | Software › Frontend & design | 83.7 | 85.9 | +2.2 |
| LMArena Agent Arena Steering Burden | Tool & computer use › General agents | 4.8 | 0.8 | −2.0 |
| Artificial Analysis Progress Index v1 | Composite indexes | 329 | 324 | −1.5 |
| LMArena Agent Arena Praise vs Complaint | Tool & computer use › General agents | -2.7 | 0.1 | +1.4 |
| τ³-Banking (Mercor run, bm25_grep retrieval) | Tool & computer use › Customer service & operations | 53.3 | 52.2 | −1.1 |
| Tax Agent Bench | Professional › Finance, accounting & tax | 71.4 | 72.4 | +1.1 |
| Artificial Analysis Intelligence Index v5.0.1 | Composite indexes | 36.0 | 36.6 | +0.6 |
| DeepFinance | Professional › Finance, accounting & tax | 36.5 | 37.0 | +0.5 |
| LMArena Agent Arena Net Improvement (Steering Burden) | Tool & computer use › General agents | 1.0 | 1.1 | +0.3 |
| Artificial Analysis Coding Agent Index v2.0 | Software › Software engineering | 45.9 | 45.7 | −0.2 |
| EnterpriseBench: CoreCraft | Tool & computer use › Customer service & operations | 62.1 | 61.9 | −0.2 |
| LMArena Agent Arena Tool Hallucination | Tool & computer use › Tool calling | 0.9 | 1.0 | +0.1 |
| RiemannBench (November 2026 rescoring) | Math | 53.6 | 53.6 | 0.0 |
Data: Artificial Analysis, Surge AI, Arena and 11 othersatlas.village.dev