Atlas

Compare Models

Kimi K3.1 Pro vs Muse Spark 1.3

Kimi K3.1 Pro leads by 2.7 capability points (≈95% interval +0.5 to +4.9): a clear lead over Muse Spark 1.3 · P(higher) ≈ 99%

Wins 53 of 77 shared benchmarks (23 lost, 1 tied)

Capability vs. relative cost
Kimi K3.1 ProMuse 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.

Kimi K3.1 ProvsMuse Spark 1.3capability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins53 – 23 · 77 shared, 1 tied

Comparing matching setups only (46 benchmarks excluded) · suites counted once
← Kimi K3.1 ProMuse Spark 1.3 →

Kimi K3.1 ProvsMuse Spark 1.3wins by categoryData Nov 15, 2026Data: Vals AI, Arena, Surge AI and 25 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 Kimi K3.1 Pro.

BrokenArXiv 08/2026 (Native Harnesses)Math64.320.2−44.1
Terminal-Bench 4.0Code63.624.7−38.9
Legal Research Bench Civil LitigationLegal29.663.0+33.3
ProofBench v1.1Math91.058.0−33.0
Legal Research Bench Business/CommercialLegal27.958.1+30.2
Legal Research Bench FamilyLegal13.640.9+27.3
Legal Research Bench ImmigrationLegal18.245.5+27.3
IOI (Vals v2)Code81.956.6−25.3
SRE Bench — Fully SolvedCode27.13.8−23.3
Terminal-Bench Science 0.1 (Vals)Science31.410.0−21.4
FrontierMath Tier 4 (v2) - PrivateMath65.946.3−19.5
ArXivMath 08/2026 (Native Harnesses)Math64.945.6−19.3
Legal Research Bench CriminalLegal28.147.4+19.3
Legal Research Bench Constitutional/Civil RightsLegal31.650.0+18.4
Mystery Game PuzzlesGames42.025.0−17.0
GameBench 2Games84.068.5−15.5
AutomationBench-AA 1.0.6Agents73.357.9−15.4
Legal Research Bench Weighted PassLegal71.486.8+15.4
MineBench — Bradley-Terry RatingGames19921747−14.8
APEX-Agents 1.1Professional Work75.961.3−14.6
FrontierML ($200 wallet)Agents29.616.0−13.6
micro1 PharmaBenchHealth & Medicine27.513.9−13.6
Creative Writing v3 EloChat & Writing21731907−13.4
BuseyBench SVGMultimodal6.95.7−12.3
Artificial Analysis Coding Agent Index v2.1Indexes48.636.8−11.9
HoneyBench v0.2 (hack rate)Safety71.883.6−11.8
Visa Agentic Commerce BenchmarkAgents53.364.4+11.1
DAYJOB: Finance (October 2026 rescoring)Finance & Accounting17.027.0+10.0
MysteryMechanismScience26.136.0+9.9
Tax Agent BenchFinance & Accounting63.972.4+8.6
Chess Puzzles (task version 2.0)Games45.037.0−8.0
Humanity's Last ExamGeneral QA51.143.2−7.9
DeepSWE v1.1Code75.567.8−7.7
τ³-Banking (Mercor run, bm25_grep retrieval)Agents59.852.2−7.6
GDPval-AA v2.1Professional Work18461681−7.3
RiemannBench (November 2026 rescoring)Math60.853.6−7.2
Humanity's Last Exam DiamondGeneral QA39.433.7−5.7
GBENCH One-shot Coding GScoreGames58.252.7−5.5
ALE-BenchCode21321981−5.3
AA-Briefcase v1.1 EloAgents16781581−5.3
LMArena Agent Arena Net Improvement (Steering Burden)Agents4.92.2−5.1
Creative Writing v3 Rubric ScoreChat & Writing16.615.6−4.9
Legal Research Bench Administrative/RegulatoryLegal56.160.6+4.5
DAYJOB: AccountingFinance & Accounting13.618.0+4.4
LMArena Code Arena WebDevCode16881655−4.3
EnterpriseBench: CoreCraftProfessional Work66.161.9−4.2
sudo L7 (October 2026 rescoring)Code16.712.8−3.9
FrontierMath Tiers 1-3 (v2) - PrivateMath78.274.4−3.9
LMCAGeneral QA57.653.9−3.7
Bug Hunt BenchCode34.330.7−3.6
Epoch Capabilities Index (ECI)Indexes162158−3.3
DTBenchGeneral QA93.396.5+3.2
LMArena Agent Arena Praise vs ComplaintAgents8.01.7−3.1
Longform Creative WritingChat & Writing79.982.8+2.9
ErdosBenchMath3.02.9−2.7
Access to Justice BenchmarkLegal69.572.2+2.7
LMArena Agent Arena Bash ReliabilityAgents7.22.0−2.6
APEX-Accounting 1.1Finance & Accounting62.464.9+2.5
NoCoT-Bench NCRI 15.2General QA10398.3−2.5
NoCoT-Bench NCKI (kspine v3)General QA116113−2.5
ACCoRDGeneral QA77.575.1−2.4
DeepFinanceFinance & Accounting34.637.0+2.4
LMArena Text Arena EloChat & Writing14831496+2.3
Conceptual Reasoning IndexIndexes67.866.2−2.3
LMArena Agent Arena Steering BurdenAgents6.72.6−2.1
MazeBench ASCIIGames2.00.0−2.0
LiveBench 2.0Agents76.774.8−2.0
DAYJOB: Data ScienceProfessional Work4.76.0+1.3
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Legal13.815.0+1.2
DAYJOB: LawLegal23.922.8−1.1
DAXBench v2Code86.285.1−1.1
Legacy-Bench (Fireworks, avg@3)Code63.062.0−1.0
Vibe Code Bench v1.1Code84.985.9+1.0
OTIS Mock AIME 2024-2025Math100.099.2−0.8
LMArena Agent Arena Confirmed SuccessAgents2.73.7+0.5
LMArena Agent Arena Tool HallucinationAgents1.00.8−0.1
Legal Research Bench HealthLegal73.373.30.0

Kimi K3.1 ProvsMuse Spark 1.3normalised score marginsData Nov 15, 2026Data: Vals AI, Arena, Surge AI and 25 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…