Atlas

Compare Models

GPT-6.2 Sol vs Muse Wonder

GPT-6.2 Sol leads by 3.5 capability points (≈95% interval +1.0 to +6.0): a clear lead over Muse Wonder · P(higher) ≈ 100%

Wins 50 of 72 shared benchmarks (22 lost)

Capability vs. relative cost
GPT-6.2 SolMuse WonderClosed weightsOpen weights

Benchmark wins50 – 22 · 72 shared

Comparing matching setups only (36 benchmarks excluded) · suites counted once
← GPT-6.2 SolMuse Wonder →
Data: Artificial Analysis, Vals AI, Surge AI and 11 othersatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Muse Wonder minus GPT-6.2 Sol.

AA-PersonalityBench Informativeness indexLanguage & writing › Style & personality37.794.3+57.1
AA-PersonalityBench Organization indexLanguage & writing › Style & personality45.790.5+45.2
Terminal-Bench-Science 0.1Science › Multidisciplinary67.625.7−41.9
KradleBenchGames67.927.6−40.4
AA-PersonalityBench Warmth indexLanguage & writing › Style & personality60.1100+40.3
AA-PersonalityBench Slop indexLanguage & writing › Style & personality15.454.3+39.3
Terminal-Bench Science 0.1 (Vals)Science › Multidisciplinary58.622.9−35.7
AA-PersonalityBench Expressiveness indexLanguage & writing › Style & personality55.388.6+33.6
AA-PersonalityBench Imagination indexLanguage & writing › Style & personality60.190.5+30.7
Blueprint-Bench 2Perception & media › Spatial & visual reasoning0.60.3−25.3
EnterpriseBench: CoreCraftTool & computer use › Customer service & operations36.461.5+25.1
Humanity's Last Exam DiamondKnowledge & reasoning › General knowledge64.039.0−25.0
LMArena Agent Arena Net Improvement (Steering Burden)Tool & computer use › General agents11.30.6−20.9
RiemannBench (November 2026 rescoring)Math80.059.2−20.8
Terminal-Bench 4.0 (Artificial Analysis)Software › Software engineering58.638.4−20.2
sudo L7 (October 2026 rescoring)Software › Software engineering36.116.1−20.0
Vals Science IndexScience › Multidisciplinary57.839.1−18.6
FrontierCode 2.0 (Artificial Analysis)Software › Software engineering44.427.9−16.5
Code Migration v1.1 (Vals Index v2.2 subset)Software › Software engineering85.369.8−15.6
Professional Reasoning Benchmark - LegalProfessional › Legal43.358.4+15.1
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & medicine37.222.2−15.0
Ramp SWE-Bench v2 (Bash harness)Software › Software engineering72.457.8−14.6
DAYJOB: Healthcare (October 2026 rescoring)Health & medicine17.231.2+14.0
Artificial Analysis Intelligence Index v5.0.1Composite indexes57.343.6−13.7
Legal Research BenchProfessional › Legal37.050.5+13.5
Humanity's Sixth SensePerception & media › Spatial & visual reasoning56.843.7−13.1
LMArena Agent Arena Praise vs ComplaintTool & computer use › General agents23.3-3.0−13.1
DAYJOB: Law (November 2026 release)Professional › Legal32.720.6−12.1
HLAB-Vals-1.1Professional › Legal12.123.8+11.7
Artificial Analysis Engineering IndexScience › Engineering61.249.6−11.6
Terminal-Bench 4.0Software › Software engineering60.649.5−11.1
BioMysteryBench (Vals Terminus 2)Science › Life sciences77.867.0−10.7
LiveBench 2.0 (2026-11-06 release)Composite indexes80.370.7−9.6
MysteryMechanismMath36.927.5−9.5
Vals Legal IndexProfessional › Legal66.775.3+8.6
Tax Agent BenchProfessional › Finance, accounting & tax62.971.2+8.3
Artificial Analysis Legal IndexProfessional › Legal55.046.9−8.2
TaxCalcBench TY25Professional › Finance, accounting & tax32.024.0−8.0
Artificial Analysis Economics IndexComposite indexes57.449.5−7.8
AA-Omniscience v1.1 IndexKnowledge & reasoning › General knowledge40.425.1−7.6
VIALS-AA (pass^3)Science › Life sciences62.755.3−7.5
CritPt v2 (agentic, avg@5)Science › Physical sciences78.671.1−7.4
Finance Agent v2.1Professional › Finance, accounting & tax58.965.5+6.6
Artificial Analysis Finance & Accounting IndexProfessional › Finance, accounting & tax57.251.1−6.1
DAYJOB: Finance (October 2026 rescoring)Professional › Finance, accounting & tax32.838.8+6.0
GDP.pdf (November 2026 rescoring)Professional › General knowledge work48.854.8+6.0
LMArena Agent Arena Steering BurdenTool & computer use › General agents13.41.5−5.9
HANDBOOK.md Agents v1.1Tool & computer use › Customer service & operations34.529.5−5.0
Artificial Analysis Strategy & Ops IndexProfessional › General knowledge work59.454.5−4.9
Agents' Last Exam v1.1 ScoreProfessional › General knowledge work73.068.4−4.7
Artificial Analysis Healthcare & Medical IndexHealth & medicine53.348.7−4.6
Vals Finance IndexProfessional › Finance, accounting & tax56.761.2+4.6
Vibe Code Bench v1.1Software › Frontend & design87.683.1−4.5
SWE Atlas v1.1 - OverallSoftware › Software engineering60.564.8+4.3
LMArena Agent Arena Confirmed SuccessTool & computer use › General agents10.92.5−4.2
VIALS-AA (average of 3)Science › Life sciences68.964.8−4.1
Dual-Arm Sim (task completion)Tool & computer use › Robotics3.80.0−3.7
DAYJOB: AccountingProfessional › Finance, accounting & tax27.630.8+3.2
AutomationBench-AA 1.0.6Tool & computer use › Customer service & operations71.368.1−3.2
AA-Briefcase v1.1 EloProfessional › General knowledge work16121659+2.6
Excel Modeling Benchmark v1.1Professional › Finance, accounting & tax74.372.5−1.8
ComplexConstraints (November 2026 rescoring)Language & writing › Instruction following69.567.9−1.6
DAYJOB: TaxProfessional › Finance, accounting & tax4.86.0+1.2
Vals Index v2.2Composite indexes66.164.9−1.2
DualEntry Accounting AI BenchmarkProfessional › Finance, accounting & tax84.283.2−1.0
AA-PersonalityBench Transparency indexLanguage & writing › Style & personality56.956.2−0.7
LMArena Agent Arena Tool HallucinationTool & computer use › Tool calling1.2-0.2−0.7
VIALS-AA (pass@3)Science › Life sciences74.573.9−0.6
GDP.pdf (Artificial Analysis, All-pass)Professional › General knowledge work30.030.6+0.6
GDP.xlsxProfessional › General knowledge work22.321.8−0.5
LMArena Agent Arena Bash ReliabilityTool & computer use › General agents4.33.4−0.4
AA-PersonalityBench Conviction indexLanguage & writing › Style & personality79.279.1−0.2
Data: Artificial Analysis, Vals AI, Surge AI and 11 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…