Hy4 vs Muse Wonder
Muse Wonder leads by 1.6 capability points (≈95% interval −1.0 to +4.1): statistically tied · P(higher) ≈ 89%
Wins 12 of 18 shared benchmarks (6 lost)
Capability vs. relative cost
Hy4Muse WonderClosed weightsOpen weights
Hy4 has no setting with a relative cost yet.
Benchmark wins6 – 12 · 18 shared
Comparing matching setups only (159 benchmarks excluded) · suites counted once
Area← Hy4Muse Wonder →
Data: Surge AI, Mercor, Neel Nanda and 6 othersatlas.village.dev
Margins
Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Muse Wonder minus Hy4.
| DAYJOB: Finance (October 2026 rescoring) | Professional › Finance, accounting & tax | 9.2 | 38.8 | +29.6 |
| DAYJOB: Healthcare (October 2026 rescoring) | Health & medicine | 7.2 | 31.2 | +24.0 |
| GDP.pdf (November 2026 rescoring) | Professional › General knowledge work | 43.6 | 54.8 | +11.2 |
| Harvey LAB (Mercor run, undisclosed ~213-task cohort) | Professional › Legal | 13.6 | 23.0 | +9.4 |
| APEX-Agents 1.1 | Professional › General knowledge work | 85.1 | 77.0 | −8.1 |
| RiemannBench (November 2026 rescoring) | Math | 51.2 | 59.2 | +8.0 |
| EnterpriseBench: CoreCraft | Tool & computer use › Customer service & operations | 68.2 | 61.5 | −6.7 |
| HANDBOOK.md Agents v1.1 | Tool & computer use › Customer service & operations | 23.7 | 29.5 | +5.8 |
| Humanity's Last Exam Diamond | Knowledge & reasoning › General knowledge | 32.7 | 37.3 | +4.6 |
| GDP.xlsx | Professional › General knowledge work | 17.7 | 21.8 | +4.1 |
| τ³-Banking (Mercor run, bm25_grep retrieval) | Tool & computer use › Customer service & operations | 48.1 | 51.9 | +3.8 |
| DAXBench v2 | Software › Data & ML | 87.2 | 84.0 | −3.2 |
| Agents' Last Exam v1.1 (ALE v1.1) | Professional › General knowledge work | 41.0 | 38.5 | −2.5 |
| NoCoT-Bench NCRI 15.2 | Knowledge & reasoning › Abstract reasoning | 94.0 | 97.7 | +2.0 |
| CWE-bench v2 Pass@1 | Software › Cybersecurity | 64.1 | 66.1 | +2.0 |
| NoCoT-Bench NCKI (kspine v3) | Knowledge & reasoning › General knowledge | 111 | 109 | −1.3 |
| LLM Debate Benchmark | Language & writing › Language & wordplay | 1647 | 1637 | −1.3 |
| ComplexConstraints (November 2026 rescoring) | Language & writing › Instruction following | 66.8 | 67.9 | +1.1 |
Data: Surge AI, Mercor, Neel Nanda and 6 othersatlas.village.dev