Atlas

Compare Models

GPT-6.2 Sol vs Muse Spark 1.3

GPT-6.2 Sol leads by 5.9 capability points (≈95% interval +3.5 to +8.3): a clear lead over Muse Spark 1.3 · P(higher) ≈ 100%

Wins 42 of 59 shared benchmarks (16 lost, 1 tied)

Capability vs. relative cost
GPT-6.2 SolMuse Spark 1.3Closed weightsOpen weights

Benchmark wins42 – 16 · 59 shared, 1 tied

Comparing matching setups only (11 benchmarks excluded) · suites counted once
← GPT-6.2 SolMuse Spark 1.3 →
Data: Artificial Analysis, Surge AI, Arena and 8 othersatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Muse Spark 1.3 minus GPT-6.2 Sol.

AA-PersonalityBench Informativeness indexLanguage & writing › Style & personality37.7100+62.9
AA-PersonalityBench Slop indexLanguage & writing › Style & personality15.475.3+60.5
CyberGym-E2E-AA (Artificial Analysis)Software › Cybersecurity0.059.5+59.5
Terminal-Bench-Science 0.1Science › Multidisciplinary67.611.0−56.7
Terminal-Bench Science 0.1 (Vals)Science › Multidisciplinary58.610.0−48.6
ChartographyPerception & media › Documents & charts75.127.6−47.5
AA-PersonalityBench Imagination indexLanguage & writing › Style & personality60.198.1+38.4
AA-PersonalityBench Warmth indexLanguage & writing › Style & personality60.196.2+36.5
DeepsecBench-AA (Artificial Analysis)Software › Cybersecurity48.912.5−36.4
Terminal-Bench 4.0Software › Software engineering60.624.7−35.9
AA-3D v1.0 EloSoftware › Frontend & design12591032−31.0
Humanity's Last Exam DiamondKnowledge & reasoning › General knowledge64.033.7−30.3
Terminal-Bench 4.0 (Artificial Analysis)Software › Software engineering60.133.3−26.8
RiemannBench (November 2026 rescoring)Math80.053.6−26.4
TaxCalcBench TY25Professional › Finance, accounting & tax32.06.0−26.0
EnterpriseBench: CoreCraftTool & computer use › Customer service & operations36.461.9+25.5
sudo L7 (October 2026 rescoring)Software › Software engineering36.112.8−23.3
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & medicine37.215.6−21.7
FrontierCode 2.0 (Artificial Analysis)Software › Software engineering44.423.1−21.3
DAYJOB: Law (November 2026 release)Professional › Legal32.711.8−20.9
Artificial Analysis Intelligence Index v5.0.1Composite indexes57.336.6−20.6
AutomationBench 1.0.6 (Zapier)Tool & computer use › Customer service & operations41.320.7−20.5
LMArena Agent Arena Net Improvement (Steering Burden)Tool & computer use › General agents11.31.1−19.9
Humanity's Sixth SensePerception & media › Spatial & visual reasoning56.837.4−19.4
AA-PersonalityBench Organization indexLanguage & writing › Style & personality45.763.8+18.3
Legal Research BenchProfessional › Legal37.055.3+18.3
Professional Reasoning Benchmark - LegalProfessional › Legal43.361.6+18.2
AA-PersonalityBench Expressiveness indexLanguage & writing › Style & personality55.371.4+16.3
AA-AnalystAgent (pass^5)Professional › General knowledge work53.837.5−16.2
AA-PersonalityBench Transparency indexLanguage & writing › Style & personality56.941.0−16.1
Gray Swan IPI Attack Success (k=15)Safety & alignment2.315.9−13.6
AutomationBench-AA 1.0.6Tool & computer use › Customer service & operations71.357.9−13.4
SimpleBenchKnowledge & reasoning › General knowledge69.981.8+11.9
LiveBench 2.0 (2026-11-06 release)Composite indexes80.368.6−11.7
LMArena Agent Arena Praise vs ComplaintTool & computer use › General agents23.30.1−11.6
CritPt v2 (agentic, avg@5)Science › Physical sciences78.668.0−10.6
DAYJOB: AccountingProfessional › Finance, accounting & tax27.618.0−9.6
Tax Agent BenchProfessional › Finance, accounting & tax62.972.4+9.5
AA-Omniscience v1.1 IndexKnowledge & reasoning › General knowledge40.423.8−8.3
Artificial Analysis Cyber Index v1Software › Cybersecurity37.744.3+6.6
LMArena Agent Arena Steering BurdenTool & computer use › General agents13.40.8−6.3
DAYJOB: Finance (October 2026 rescoring)Professional › Finance, accounting & tax32.827.0−5.8
DAYJOB: Healthcare (October 2026 rescoring)Health & medicine17.222.0+4.8
ComplexConstraints (November 2026 rescoring)Language & writing › Instruction following69.564.8−4.7
LMArena Agent Arena Confirmed SuccessTool & computer use › General agents10.92.5−4.2
GDP.pdf (November 2026 rescoring)Professional › General knowledge work48.853.0+4.2
GDP.pdf (Artificial Analysis, All-pass)Professional › General knowledge work30.026.6−3.4
CWE-Bench-AA (Artificial Analysis)Software › Cybersecurity64.260.8−3.3
Harvey LAB-AA v1.1 (hallucination-gated)Professional › Legal11.18.9−2.2
AA-PersonalityBench Conviction indexLanguage & writing › Style & personality79.277.2−2.1
Vibe Code Bench v1.1Software › Frontend & design87.685.9−1.8
AA-Briefcase v1.1 EloProfessional › General knowledge work16121581−1.7
LMArena Agent Arena Bash ReliabilityTool & computer use › General agents4.31.2−1.6
DAYJOB: TaxProfessional › Finance, accounting & tax4.83.6−1.2
MysteryMechanismMath36.936.0−0.9
HANDBOOK.md Agents v1.1Tool & computer use › Customer service & operations34.533.9−0.6
GDP.xlsxProfessional › General knowledge work22.322.6+0.3
LMArena Agent Arena Tool HallucinationTool & computer use › Tool calling1.21.0−0.1
DrugDiscoveryBenchScience › Life sciences62.262.20.0
Data: Artificial Analysis, Surge AI, Arena and 8 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…