Atlas

Compare Models

GPT-5.6 Sol vs Muse Spark 1.3

GPT-5.6 Sol leads by 1.7 capability points (≈95% interval −0.4 to +3.8): statistically tied · P(higher) ≈ 94%

Wins 85 of 152 shared benchmarks (67 lost)

Capability vs. relative cost
GPT-5.6 SolMuse Spark 1.3Closed weightsOpen weights

Each line runs through one model’s priced reasoning settings, cheapest to most expensive; grey points are other models’ priced settings in the same range. Results whose reasoning effort the source didn’t report are left out wherever a model has named effort levels. Capability and relative cost are the published estimates (relative cost is cost per task as a multiple of GPT-5 (high)); hover a setting, or use the arrow keys, to set it beside the other model’s same setting.

GPT-5.6 SolvsMuse Spark 1.3capability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins85 – 67 · 152 shared

Comparing matching setups only (72 benchmarks excluded) · suites counted once
← GPT-5.6 SolMuse Spark 1.3 →

GPT-5.6 SolvsMuse Spark 1.3wins by categoryData Nov 15, 2026Data: Vals AI, Artificial Analysis, Arena and 48 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Muse Spark 1.3 minus GPT-5.6 Sol.

BabaIsBench Stage 0: The IntroGames100.046.0−54.0
VoxelBench Text RatingGames22971616−48.5
RiemannBenchMath74.428.0−46.4
Olam Social Poker EloGames15341494−43.5
FrontierMath Tier 4 (v2) - PrivateMath82.946.3−36.6
IOI (Vals v2)Code91.256.6−34.6
HiL-Bench (Human-in-Loop Benchmark)Agents32.366.9+34.6
Mystery Game PuzzlesGames58.025.0−33.0
SRE Bench — Fully SolvedCode30.53.8−26.7
ProofBench v1.1Math83.058.0−25.0
OpenAI MRCR v2 8-needle - 512K-1MGeneral QA73.898.1+24.3
TaxCalcBench TY25Finance & Accounting30.06.0−24.0
WordleBenchGames97.073.0−24.0
ITBench-AAAgents56.233.2−23.0
Hedge-BenchFinance & Accounting31.59.1−22.5
Chess Puzzles (task version 2.0)Games59.037.0−22.0
LLM Creative Story WritingChat & Writing2.50.6−21.6
Harvey's Legal Agent BenchmarkLegal2.523.8+21.3
TextQuests Game ProgressGames52.133.0−19.1
Finance Agent v2 - General Qualitative AnalysisFinance & Accounting64.983.4+18.4
dfbench Detection RecallCybersecurity65.747.4−18.3
ChartographyMultimodal45.027.6−17.4
Medical Long Context Reasoning (MLCR-AA)Health & Medicine26.143.3+17.2
EnterpriseBench: CoreCraftProfessional Work46.261.9+15.7
ALE-BenchCode24101981−15.2
HardGeoBench <=100miMultimodal40.055.0+15.0
GBENCH One-shot Coding GScoreGames67.752.7−15.0
FrontierMath Tiers 1-3 (v2) - PrivateMath89.174.4−14.7
BrokenArXiv 06/2026Math67.353.2−14.0
AutoResearchExam (hidden-test AUARC)Agents0.50.4−13.4
Terminal-Bench 4.0Code37.924.7−13.1
Design Creativity BenchCode51.347.2−13.1
Finance Agent v2 - Financial ModelingFinance & Accounting20.733.5+12.8
HardGeoBench <=10miMultimodal26.338.8+12.5
HardGeoBench Country AccuracyMultimodal47.560.0+12.5
ArXivMath 06/2026Math88.576.7−11.8
CheatBenchSafety74.986.5−11.6
GameBench 2Games80.168.5−11.5
Terminal-Bench-Science 0.1Science22.411.0−11.4
DistressBenchSafety73.684.9+11.3
Gray Swan IPI Attack Success (k=15)Safety27.015.9+11.1
HardGeoBench Avg DistanceMultimodal13811022+11.1
AA-AnalystAgent (pass^5)Professional Work47.537.5−10.0
Terminal-Bench Science 0.1 (Vals)Science20.010.0−10.0
DAYJOB: FinanceFinance & Accounting5.514.8+9.3
CritPt v2 (agentic, avg@5)Science77.168.0−9.1
Finance Agent v2 - PrecedentsFinance & Accounting26.235.2+9.0
APEX-SWECode48.039.1−8.9
Extended NYT Connections 940 (Quadratic v1)Games93.885.1−8.7
WeirdML v3Science15.06.7−8.3
Finance Agent v2 - All PassFinance & Accounting41.449.5+8.2
HANDBOOK.mdProfessional Work23.515.4−8.1
AutomationBench 1.0.6 (Zapier)Agents28.820.7−8.1
LMArena Agent Arena Net ImprovementAgents6.33.8−8.0
LLM Debate BenchmarkChat & Writing16451593−7.5
Humanity's Sixth SenseMultimodal30.037.4+7.4
Legal Research BenchLegal48.155.3+7.2
Arena Alignment IndexSafety84.277.1−7.1
OpenAI MRCR v2 8-needle - 256K-512KGeneral QA91.598.5+7.0
Shroombench (mushroom identification, top-1)Multimodal42.049.0+7.0
Finance Agent v2 - Disclosure AnalysisFinance & Accounting63.370.1+6.7
AA-Briefcase EloAgents14751589+6.6
Terminal-Bench 4.0 (Artificial Analysis)Code39.933.3−6.6
FrontierSWE V2Code32.225.8−6.4
Document ArenaProfessional Work14801473−6.4
KernelBench Mega Normalized (H100)Code18.612.4−6.3
DAYJOB: HealthcareHealth & Medicine1.67.8+6.2
𝜏³-BankingAgents44.350.5+6.2
NoCoT-Bench NCRI 15.2General QA10998.3−6.1
Finance Agent v2 - Market AnalysisFinance & Accounting64.269.9+5.8
DrugDiscoveryBenchScience56.562.2+5.7
Code MigrationCode52.947.4−5.5
NoCoT-Bench NCKI (kspine v3)General QA121113−5.5
Vibe Code Bench v1.1Code80.585.9+5.4
Excel Modeling BenchmarkFinance & Accounting72.367.4−4.9
OSWorld 2.0 - reported binary completion, patch and step limit unspecifiedAgents27.332.0+4.7
LMArena Agent Arena Bash ReliabilityAgents11.42.0−4.7
DeepSWE v1.1Code70.775.4+4.7
LMCAGeneral QA58.553.9−4.6
LMArena Agent Arena Net Improvement (Steering Burden)Agents4.62.2−4.6
SWE-TogetherCode57.862.4+4.6
Tax Agent BenchFinance & Accounting68.072.4+4.5
HackerRank-BenchCode38.442.9+4.4
OSWorld 2.0 - reported partial reward, patch and step limit unspecifiedAgents62.766.9+4.2
Terminal-Bench 2.1Code89.585.4−4.1
Epoch Capabilities Index (ECI)Indexes163158−4.1
BuseyBench SVGMultimodal6.15.7−4.0
WellnessBenchSafety57.061.0+4.0
CadQueryEval (Corrected Geometry)Code88.084.0−4.0
Financial Coaching BenchFinance & Accounting74.971.0−3.9
GDPval-AA v2Professional Work15851662+3.8
Finance Agent v2 - AdjustmentsFinance & Accounting49.252.8+3.6
CyberBench v1.1 OverallCybersecurity76.372.7−3.6
Conceptual Reasoning IndexIndexes68.766.2−3.5
Creative Writing v3 EloChat & Writing19741907−3.4
Humanity's Last Exam DiamondGeneral QA30.433.7+3.3
GPQA DiamondGeneral QA94.691.3−3.3
GDP.pdfProfessional Work30.727.6−3.1
Finance Agent v2 - Earnings AnalysisFinance & Accounting74.477.2+2.8
DeepSearchQA (F1 score)Search93.190.3−2.8
ObviousBenchGeneral QA96.599.3+2.8
GDPval-AA v2.1Professional Work16191681+2.7
MysteryMechanismScience33.336.0+2.7
LMArena Agent Arena Bash RecoveryAgents0.76.1+2.7
Creative Writing v3 Rubric ScoreChat & Writing16.215.6−2.7
CritPt (avg@5)Science28.626.0−2.6
LMArena Agent Arena SteerabilityAgents6.51.6−2.5
Artificial Analysis Progress Index v1Indexes331324−2.5
Finance Agent v2 - General Quantitative AnalysisFinance & Accounting75.678.0+2.4
Humanity's Last ExamGeneral QA45.543.2−2.3
LMArena Agent Arena Praise vs ComplaintAgents6.31.7−2.3
AutomationBench-AA 1.0.6Agents60.157.9−2.2
ChessBench CoherenceGames89.191.2+2.1
AURORA Multilingual Terminal-BenchCode64.562.5−2.0
APEX-AccountingFinance & Accounting51.553.5+2.0
GeneBench-Pro Public Release SubsetScience24.026.0+2.0
LMArena Text Arena EloChat & Writing14851496+2.0
SciCode 1.0.1 (Artificial Analysis)Code57.859.7+2.0
Finance Agent v2 - ComparablesFinance & Accounting45.347.2+1.9
Design Arena Overall Code CategoriesCode13381348+1.7
Agents' Last Exam (ALE)Agents30.632.2+1.6
AA-Omniscience IndexIndexes22.025.0+1.5
LiveBench 2.0Agents76.374.8−1.5
LMArena Agent Arena Steering BurdenAgents5.52.6−1.5
ComplexConstraintsChat & Writing50.551.9+1.4
Artificial Analysis Intelligence Index v4.3.2Indexes47.048.1+1.1
Artificial Analysis Intelligence Index v4.3Indexes47.148.2+1.1
CVDP (517-problem set)Code55.454.3−1.1
Realm Financial ReasoningFinance & Accounting44.445.5+1.1
Longform Creative WritingChat & Writing81.782.8+1.1
Ruby LLM Benchmarks — Program FixerCode75.274.1−1.1
PCT Political Inconsistency (CAIS)Safety41.740.6+1.1
ChessBench EloGames12181201−1.0
AA-LCR v1.1Search84.083.0−1.0
Eval Connections One-Shot with TrapsGames83.084.0+1.0
ProgramBenchCode1.52.5+1.0
Vision Arena OverallMultimodal12881291+0.9
OTIS Mock AIME 2024-2025Math100.099.2−0.8
ACCoRDGeneral QA75.975.1−0.8
Vals Index v2Indexes63.764.5+0.8
DTBenchGeneral QA95.796.5+0.8
Roboflow Vision Evals (September 2026 protocol)Multimodal79.079.8+0.8
LMArena Agent Arena Confirmed SuccessAgents2.33.7+0.7
ForecastBench Preliminary Dataset Brier IndexForecasting61.461.9+0.5
ChessBench AccuracyGames76.476.0−0.4
Bench to the Future 3 — September Panel — PooledForecasting0.10.1−0.4
SpeciEval (Refusal-Aware Retries, Historical)Safety99.098.8−0.3
SpeciEval (Pooled English Runs)Safety99.098.7−0.3
ForecastBench Tournament Overall Brier IndexForecasting66.165.9−0.2
Vals Index v2.1Indexes58.058.2+0.2
LMArena Agent Arena Tool HallucinationAgents1.00.8−0.1
CursorBench 4.0Code41.741.6−0.1

GPT-5.6 SolvsMuse Spark 1.3normalised score marginsData Nov 15, 2026Data: Vals AI, Artificial Analysis, Arena and 48 othersVillage LLM Atlasatlas.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…