Claude Haiku 5.5 vs Muse Ember
TryTop 2 now
Muse Ember leads by 0.9 capability points (≈95% interval −1.3 to +3.2): statistically tied · P(higher) ≈ 79%
Wins 29 of 51 shared benchmarks (21 lost, 1 tied)
Capability vs. relative cost
Claude Haiku 5.5Muse EmberClosed weightsOpen weights
Benchmark wins21 – 29 · 51 shared, 1 tied
Comparing matching setups only (1 benchmark excluded) · suites counted once
Area← Claude Haiku 5.5Muse 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 Claude Haiku 5.5.
| AutomationBench-AA 1.0.6 | Tool & computer use › Customer service & operations | 36.0 | 68.1 | +32.1 |
| HLAB-Vals-1.1 | Professional › Legal | 4.2 | 25.0 | +20.8 |
| DAXBench v2 | Software › Data & ML | 81.9 | 62.8 | −19.1 |
| Harvey LAB (Mercor run, undisclosed ~213-task cohort) | Professional › Legal | 4.9 | 22.7 | +17.8 |
| Medical Long Context Reasoning (MLCR-AA, October 2026 methodology) | Health & medicine | 35.0 | 18.3 | −16.7 |
| SimpleBench | Knowledge & reasoning › General knowledge | 59.2 | 72.7 | +13.5 |
| KradleBench | Games | 15.0 | 25.7 | +10.7 |
| τ³-Banking (Mercor run, bm25_grep retrieval) | Tool & computer use › Customer service & operations | 43.0 | 53.3 | +10.3 |
| Artificial Analysis Strategy & Ops Index | Professional › General knowledge work | 39.1 | 48.7 | +9.6 |
| Tax Agent Bench | Professional › Finance, accounting & tax | 62.2 | 71.4 | +9.1 |
| BuseyBench SVG | Perception & media › Generation | 4.6 | 5.4 | +8.4 |
| Terminal-Bench 4.0 (Artificial Analysis) | Software › Software engineering | 32.8 | 40.9 | +8.1 |
| FrontierCode 2.0 (Artificial Analysis) | Software › Software engineering | 34.9 | 27.2 | −7.8 |
| APEX-Agents 1.1 | Professional › General knowledge work | 68.3 | 75.7 | +7.4 |
| Vibe Code Bench v1.1 | Software › Frontend & design | 90.4 | 83.7 | −6.7 |
| Excel Modeling Benchmark v1.1 | Professional › Finance, accounting & tax | 65.5 | 72.2 | +6.7 |
| APEX-1 (Mercor, mean score) | Professional › General knowledge work | 64.7 | 70.4 | +5.7 |
| Ramp SWE-Bench v2 | Software › Software engineering | 57.4 | 51.8 | −5.6 |
| Vals Legal Index | Professional › Legal | 70.3 | 75.7 | +5.4 |
| Legal Research Bench | Professional › Legal | 43.8 | 49.0 | +5.3 |
| Humanity's Last Exam Diamond | Knowledge & reasoning › General knowledge | 29.6 | 25.1 | −4.5 |
| CritPt v2 (agentic, avg@5) | Science › Physical sciences | 66.3 | 62.0 | −4.3 |
| Artificial Analysis Healthcare & Medical Index | Health & medicine | 37.5 | 41.3 | +3.8 |
| DeepFinance | Professional › Finance, accounting & tax | 40.0 | 36.5 | −3.5 |
| Vals Finance Index | Professional › Finance, accounting & tax | 57.4 | 60.9 | +3.5 |
| SWE Atlas v1.1 - Overall | Software › Software engineering | 59.4 | 56.0 | −3.4 |
| Code Migration v1.1 (Vals Index v2.2 subset) | Software › Software engineering | 50.2 | 47.1 | −3.1 |
| LMArena Agent Arena Net Improvement (Steering Burden) | Tool & computer use › General agents | -0.6 | 1.0 | +3.1 |
| Terminal-Bench 4.0 | Software › Software engineering | 36.4 | 39.4 | +3.0 |
| AA-Briefcase v1.1 Elo | Professional › General knowledge work | 1572 | 1520 | −2.8 |
| FORTRESS Average Risk Score (ARS) | Safety & alignment | 9.5 | 11.9 | −2.4 |
| AA-Omniscience v1.1 Index | Knowledge & reasoning › General knowledge | 12.3 | 7.7 | −2.3 |
| APEX-SWE | Software › Software engineering | 55.1 | 52.9 | −2.2 |
| LMArena Agent Arena Steering Burden | Tool & computer use › General agents | 0.8 | 4.8 | +2.0 |
| GDP.pdf (Artificial Analysis, All-pass) | Professional › General knowledge work | 20.8 | 22.8 | +2.0 |
| TaxCalcBench TY25 | Professional › Finance, accounting & tax | 4.0 | 2.0 | −2.0 |
| Vals Index v2.2 | Composite indexes | 58.7 | 60.5 | +1.7 |
| Artificial Analysis Economics Index | Composite indexes | 42.7 | 41.0 | −1.7 |
| LMArena Code Arena WebDev | Software › Frontend & design | 1596 | 1608 | +1.7 |
| Finance Agent v2.1 | Professional › Finance, accounting & tax | 65.4 | 63.8 | −1.5 |
| LMArena Agent Arena Praise vs Complaint | Tool & computer use › General agents | -5.5 | -2.7 | +1.4 |
| LMArena Agent Arena Tool Hallucination | Tool & computer use › Tool calling | -1.5 | 0.9 | +1.2 |
| LMArena Agent Arena Confirmed Success | Tool & computer use › General agents | 0.2 | -2.0 | −1.1 |
| GBENCH One-shot Coding GScore | Games | 55.4 | 56.4 | +1.0 |
| Artificial Analysis Finance & Accounting Index | Professional › Finance, accounting & tax | 43.6 | 44.6 | +1.0 |
| Artificial Analysis Progress Index v1 | Composite indexes | 326 | 329 | +0.9 |
| Artificial Analysis Engineering Index | Science › Engineering | 42.5 | 41.8 | −0.8 |
| Artificial Analysis Intelligence Index v5.0.1 | Composite indexes | 36.5 | 36.0 | −0.5 |
| Artificial Analysis Legal Index | Professional › Legal | 38.7 | 38.5 | −0.2 |
| LMArena Agent Arena Bash Reliability | Tool & computer use › General agents | 6.6 | 6.6 | 0.0 |
| Terminal-Bench-Science 0.1 | Science › Multidisciplinary | 20.0 | 20.0 | 0.0 |
Data: Artificial Analysis, Vals AI, Arena and 10 othersatlas.village.dev