Atlas

Compare Models

GPT-5.5 vs Muse Spark 1.3

Muse Spark 1.3 leads by 1.2 capability points (≈95% interval −0.9 to +3.2): statistically tied · P(higher) ≈ 87%

Wins 70 of 131 shared benchmarks (58 lost, 3 tied)

Capability vs. relative cost
GPT-5.5Muse 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-5.5vsMuse Spark 1.3capability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins58 – 70 · 131 shared, 3 tied

Comparing matching setups only (61 benchmarks excluded) · suites counted once
← GPT-5.5Muse Spark 1.3 →

GPT-5.5vsMuse Spark 1.3wins by categoryData Nov 15, 2026Data: Artificial Analysis, Vals AI, Mercor and 39 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-5.5.

BabaIsBench Stage 0: The IntroGames100.046.0−54.0
Mystery Game PuzzlesGames56.025.0−31.0
Olam Social Poker EloGames15211494−29.3
HiL-Bench (Human-in-Loop Benchmark)Agents39.766.9+27.2
RiemannBenchMath55.228.0−27.2
AutomationBench (Public Set)Agents22.749.6+26.9
JobBench MainProfessional Work38.364.9+26.6
FrontierMath Tier 4 (v2) - PrivateMath72.546.3−26.2
Chess Puzzles (task version 2.0)Games63.037.0−26.0
Medical Long Context Reasoning (MLCR-AA)Health & Medicine17.843.3+25.6
OpenAI MRCR v2 8-needle - 512K-1MGeneral QA74.098.1+24.1
WordleBenchGames97.073.0−24.0
BrokenArXiv 04/2026Math72.149.2−22.9
VoxelBench Text RatingGames19251616−22.0
LLM Creative Story WritingChat & Writing2.40.6−21.2
AA-Briefcase EloAgents11361483+20.2
Harvey's Legal Agent BenchmarkLegal3.823.8+20.0
GeneBench-Pro Public Release SubsetScience8.026.0+18.0
TaxCalcBench TY25Finance & Accounting24.06.0−18.0
Finance Agent v2 - General Qualitative AnalysisFinance & Accounting65.683.4+17.8
Code MigrationCode45.227.6−17.6
OpenAI MRCR v2 8-needle - 256K-512KGeneral QA81.598.5+17.0
BrokenArXiv 06/2026Math69.453.2−16.2
TextQuests Game ProgressGames48.933.0−15.9
HardGeoBench <=100miMultimodal40.055.0+15.0
HardGeoBench Country AccuracyMultimodal45.060.0+15.0
Legal Research BenchLegal40.455.3+14.9
SWE-bench Verified Extended (Mercor held-out)Code39.354.0+14.7
GDPval-AA v2Professional Work13721662+14.3
BrokenArXiv 05/2026Math50.036.5−13.5
Finance Agent v2 - PrecedentsFinance & Accounting21.935.2+13.3
Vibe Code Bench v1.1Code69.882.9+13.0
Tax Agent BenchFinance & Accounting60.573.4+12.9
SimpleBenchGeneral QA69.081.8+12.8
ITBench-AAAgents45.833.2−12.6
AA-AnalystAgent (pass^5)Professional Work50.037.5−12.5
GDPval-AA v2.1Professional Work13571631+12.1
𝜏³-BankingAgents39.050.5+11.5
Finance Agent v2 - Disclosure AnalysisFinance & Accounting58.970.1+11.2
Extended NYT Connections 940 (Quadratic v1)Games96.285.1−11.1
DeepSWE v1.1Code64.475.4+11.0
FrontierMath Tiers 1-3 (v2) - PrivateMath85.374.4−10.9
KernelBench Mega Normalized (H100)Code23.112.4−10.8
DrugDiscoveryBenchScience51.662.2+10.6
EnterpriseBench: CoreCraftProfessional Work51.361.9+10.6
Finance Agent v2 - Financial ModelingFinance & Accounting22.933.5+10.6
AutomationBench-AA 1.0.6Agents47.357.9+10.6
Document ArenaProfessional Work14841473−10.2
HardGeoBench <=10miMultimodal28.838.8+10.0
MMMU-Pro Extended (Mercor held-out)Multimodal43.833.8−10.0
Finance Agent v2 - All PassFinance & Accounting39.649.5+10.0
Artificial Analysis Intelligence Index v4.3.2Indexes38.448.1+9.7
Finance Agent v2 - Market AnalysisFinance & Accounting60.369.9+9.7
Artificial Analysis Intelligence Index v4.3Indexes38.648.2+9.5
GBENCH One-shot Coding GScoreGames62.252.7−9.5
LLM Debate BenchmarkChat & Writing15281593+9.3
ChessBench EloGames13491201−8.8
ChessBench CoherenceGames100.091.2−8.8
GameBench 2Games59.968.5+8.6
CadQueryEval (Corrected Geometry)Code76.084.0+8.0
Agents' Last Exam (ALE)Agents24.532.2+7.8
Vals Index v2Indexes57.464.5+7.1
Finance Agent v2 - AdjustmentsFinance & Accounting45.752.8+7.1
ArXivMath 06/2026Math83.676.7−6.9
HANDBOOK.mdProfessional Work21.515.4−6.1
LiveBench 2.0Agents68.774.8+6.1
MineBench — Bradley-Terry RatingGames18461747−6.0
Humanity's Last Exam DiamondGeneral QA27.933.7+5.8
APEX-1 (Mercor, mean score)Professional Work68.473.9+5.5
ACCoRDGeneral QA80.575.1−5.4
HardGeoBench Avg DistanceMultimodal11961022+5.4
APEX-AccountingFinance & Accounting48.253.5+5.3
Roboflow Vision Evals (September 2026 protocol)Multimodal74.879.8+5.0
Longform Creative WritingChat & Writing78.282.8+4.6
ALE-BenchCode21081981−4.5
Finance Agent v2 - General Quantitative AnalysisFinance & Accounting73.778.0+4.3
ArXivMath 05/2026Math77.573.3−4.2
AutomationBench 1.0.6 (Zapier)Agents16.620.7+4.1
Finance Agent v2 - Earnings AnalysisFinance & Accounting73.277.2+4.0
AA-LCR Extended (Mercor held-out)Search70.566.9−3.6
SciCode 1.0.1 (Artificial Analysis)Code56.159.7+3.6
Finance Agent v2 - ComparablesFinance & Accounting43.747.2+3.5
Ruby LLM Benchmarks — Program FixerCode77.674.1−3.5
ChartographyMultimodal31.027.6−3.4
DAXBench v2Code88.385.1−3.2
Creative Writing v3 EloChat & Writing18451907+3.2
ChessBench AccuracyGames78.876.0−3.0
BuseyBench SVGMultimodal6.05.7−2.9
Excel Modeling BenchmarkFinance & Accounting64.567.4+2.9
LMArena Agent Arena SteerabilityAgents6.91.6−2.6
LMArena Text Arena EloChat & Writing14811496+2.6
DeepSearchQA (F1 score)Search87.890.3+2.5
ComplexConstraintsChat & Writing49.551.9+2.4
AA-Omniscience IndexIndexes20.525.0+2.2
Vision Arena OverallMultimodal12831291+2.1
ArXivMath 04/2026Math68.866.7−2.1
Terminal-Bench 4.0 (Artificial Analysis)Code14.616.7+2.0
LMArena Agent Arena Bash RecoveryAgents9.56.1−1.7
Epoch Capabilities Index (ECI)Indexes160158−1.7
GDP.pdfProfessional Work26.027.6+1.6
Conceptual Reasoning IndexIndexes67.366.2−1.6
Artificial Analysis Progress Index v1Indexes320324+1.6
Code2Video-AA EloCode10001019+1.5
ForecastBench Preliminary Dataset Brier IndexForecasting63.361.9−1.4
GDPval Extended (Mercor held-out rubric score)Professional Work54.555.9+1.4
NoCoT-Bench NCKI (kspine v3)General QA111113+1.4
LMArena Agent Arena Net ImprovementAgents4.23.8−1.4
AA-LCR v1.1Search84.383.0−1.3
Creative Writing v3 Rubric ScoreChat & Writing15.915.6−1.2
Terminal-Bench 2.1Code84.385.4+1.1
ForecastBench Tournament Overall Brier IndexForecasting67.065.9−1.1
NoCoT-Bench NCRI 15.2General QA10098.3−1.0
CharXiv Extended (Mercor held-out)Multimodal82.781.8−0.9
OTIS Mock AIME 2024-2025Math100.099.2−0.8
MedXpertQA MM Extended (Mercor held-out)Health & Medicine60.860.1−0.7
CharXiv (descriptive and reasoning combined)Multimodal94.693.9−0.7
ObviousBenchGeneral QA100.099.3−0.7
PCT Political Inconsistency (CAIS)Safety40.040.6−0.6
GPQA Diamond (Artificial Analysis avg@5)General QA93.594.1+0.6
Humanity's Last ExamGeneral QA43.843.2−0.6
CritPt (avg@5)Science25.426.0+0.6
SpeciEval (Refusal-Aware Retries, Historical)Safety98.298.8+0.6
SpeciEval (Pooled English Runs)Safety98.298.7+0.6
DTBenchGeneral QA96.096.5+0.5
LMCAGeneral QA54.353.9−0.4
MedXpertQA MM (200-question set)Health & Medicine76.776.3−0.4
APEX-SWECode36.636.4−0.2
APEX-Agents 1.1Professional Work70.370.4+0.1
AIME 2025Math100.0100.00.0
ProgramBenchCode0.50.50.0
SRE Bench — Fully SolvedCode3.83.80.0

GPT-5.5vsMuse Spark 1.3normalised score marginsData Nov 15, 2026Data: Artificial Analysis, Vals AI, Mercor and 39 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…