Atlas

Compare Models

GPT-6.1 Sol vs Muse Ember

GPT-6.1 Sol leads by 4.0 capability points (≈95% interval +1.5 to +6.5): a clear lead over Muse Ember · P(higher) ≈ 100%

Wins 33 of 53 shared benchmarks (20 lost)

Capability vs. relative cost
GPT-6.1 SolMuse EmberClosed weightsOpen weights

Benchmark wins33 – 20 · 53 shared

Comparing matching setups only (11 benchmarks excluded) · suites counted once
← GPT-6.1 SolMuse Ember →
Data: Artificial Analysis, Vals AI, Surge AI and 10 othersatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Muse Ember minus GPT-6.1 Sol.

Terminal-Bench-Science 0.1Science › Multidisciplinary58.120.0−38.1
TaxCalcBench TY25Professional › Finance, accounting & tax36.02.0−34.0
Humanity's Last Exam DiamondKnowledge & reasoning › General knowledge58.825.1−33.7
DAXBench v2Software › Data & ML92.662.8−29.8
KradleBenchGames54.625.7−28.9
Code Migration v1.1 (Vals Index v2.2 subset)Software › Software engineering75.147.1−28.0
RiemannBench (November 2026 rescoring)Math79.253.6−25.6
BuseyBench SVGPerception & media › Generation8.05.4−25.5
GBENCH One-shot Coding GScoreGames77.556.4−21.1
CritPt v2 (agentic, avg@5)Science › Physical sciences82.662.3−20.3
LMArena Agent Arena Net Improvement (Steering Burden)Tool & computer use › General agents11.31.0−20.2
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & medicine38.318.3−20.0
HLAB-Vals-1.1Professional › Legal6.325.0+18.8
Artificial Analysis Intelligence Index v5.0.1Composite indexes54.636.0−18.6
AA-Omniscience v1.1 IndexKnowledge & reasoning › General knowledge42.47.7−17.3
Terminal-Bench 4.0Software › Software engineering55.139.4−15.7
Terminal-Bench 4.0 (Artificial Analysis)Software › Software engineering56.140.9−15.2
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Professional › Legal8.822.7+13.9
LMArena Agent Arena Praise vs ComplaintTool & computer use › General agents24.5-2.7−13.6
FrontierCode 2.0 (Artificial Analysis)Software › Software engineering40.627.2−13.5
EnterpriseBench: CoreCraftTool & computer use › Customer service & operations50.362.1+11.8
HiL-Bench (Human-in-Loop Benchmark)Tool & computer use › General agents62.374.0+11.7
ComplexConstraints (November 2026 rescoring)Language & writing › Instruction following70.158.8−11.3
Legal Research BenchProfessional › Legal38.549.0+10.6
τ³-Banking (Mercor run, bm25_grep retrieval)Tool & computer use › Customer service & operations43.053.3+10.3
SimpleBenchKnowledge & reasoning › General knowledge82.972.7−10.2
SWE Atlas v1.1 - OverallSoftware › Software engineering65.956.0−9.9
Professional Reasoning Benchmark - FinanceProfessional › Finance, accounting & tax46.256.0+9.8
DAYJOB: Finance (October 2026 rescoring)Professional › Finance, accounting & tax33.524.2−9.3
GDP.pdf (Artificial Analysis, All-pass)Professional › General knowledge work32.022.8−9.2
Tax Agent BenchProfessional › Finance, accounting & tax62.371.4+9.0
DAYJOB: Healthcare (October 2026 rescoring)Health & medicine25.217.2−8.0
Professional Reasoning Benchmark - LegalProfessional › Legal47.955.6+7.7
Vals Legal IndexProfessional › Legal68.775.7+7.0
Artificial Analysis Progress Index v1Composite indexes348329−6.9
LMArena Agent Arena Confirmed SuccessTool & computer use › General agents11.6-2.0−6.8
GDP.xlsxProfessional › General knowledge work22.015.4−6.6
APEX-SWESoftware › Software engineering46.952.9+6.0
Vibe Code Bench v1.1Software › Frontend & design88.983.7−5.2
Vals Finance IndexProfessional › Finance, accounting & tax55.860.9+5.1
LMArena Agent Arena Steering BurdenTool & computer use › General agents11.84.8−3.5
Vals Index v2.2Composite indexes63.960.5−3.5
FORTRESS Average Risk Score (ARS)Safety & alignment14.711.9+2.8
APEX-Agents 1.1Professional › General knowledge work73.275.7+2.5
Ramp SWE-Bench v2 (Bash harness)Software › Software engineering52.054.4+2.4
GDP.pdf (November 2026 rescoring)Professional › General knowledge work51.649.2−2.4
Finance Agent v2.1Professional › Finance, accounting & tax61.863.8+2.0
AA-Briefcase v1.1 EloProfessional › General knowledge work15571520−2.0
AutomationBench-AA 1.0.6Tool & computer use › Customer service & operations66.668.1+1.5
HANDBOOK.md Agents v1.1Tool & computer use › Customer service & operations28.629.8+1.2
Excel Modeling Benchmark v1.1Professional › Finance, accounting & tax71.172.2+1.1
LMArena Agent Arena Bash ReliabilityTool & computer use › General agents5.16.6+0.8
LMArena Agent Arena Tool HallucinationTool & computer use › Tool calling1.20.9−0.1
Data: Artificial Analysis, Vals AI, Surge AI and 10 othersatlas.village.dev
Win counts use each benchmark’s declared score direction. Represented suite slices and the derived Capability score are excluded from headline totals. Missing or incompatible setups are excluded, never counted as losses.
Loading Atlas data…