Atlas

Compare Models

GPT-5.5 vs Kimi K3.1 Pro

Kimi K3.1 Pro leads by 3.8 capability points (≈95% interval +1.7 to +6.0): a clear lead over GPT-5.5 · P(higher) ≈ 100%

Wins 45 of 66 shared benchmarks (19 lost, 2 tied)

Capability vs. relative cost
GPT-5.5Kimi K3.1 ProClosed 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-5.5vsKimi K3.1 Procapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins19 – 45 · 66 shared, 2 tied

Comparing matching setups only (39 benchmarks excluded) · suites counted once
← GPT-5.5Kimi K3.1 Pro →

GPT-5.5vsKimi K3.1 Prowins by categoryData Nov 15, 2026Data: Vals AI, SecureBio, Epoch AI and 21 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Kimi K3.1 Pro minus GPT-5.5.

AA-Briefcase v1.1 EloAgents11351678+29.8
AutomationBench-AA 1.0.6Agents47.373.3+26.0
GameBench 2Games59.984.0+24.1
SRE Bench — Fully SolvedCode3.827.1+23.3
ZeroBench (Main questions, pass@5, with tools)Multimodal41.063.0+22.0
GDPval-AA v2.1Professional Work13571846+21.5
Legal Research Bench ImmigrationLegal36.418.2−18.2
Chess Puzzles (task version 2.0)Games63.045.0−18.0
Creative Writing v3 EloChat & Writing18452173+16.6
DeepsecBench — Revalidated V1 (233 issues)Code21.037.5+16.4
Design Arena Overall Code CategoriesCode12571353+15.9
Vibe Code Bench v1.1Code69.884.9+15.0
Legal Research Bench Civil LitigationLegal44.429.6−14.8
EnterpriseBench: CoreCraftProfessional Work51.366.1+14.8
Mystery Game PuzzlesGames56.042.0−14.0
OfficeQA ProProfessional Work54.167.4+13.3
ZeroBench Main Mean pass@1 (five samples)Multimodal17.229.8+12.6
Legal Research Bench Weighted PassLegal84.071.4−12.5
PerceptionBenchMultimodal55.868.3+12.5
BabyVision with PythonMultimodal83.695.4+11.8
Furniture Assembly BenchmarkMultimodal44.255.8+11.7
Humanity's Last Exam DiamondGeneral QA27.939.4+11.5
DeepSWE v1.1Code64.475.5+11.1
BuseyBench SVGMultimodal6.06.9+9.4
Legal Research Bench Business/CommercialLegal37.227.9−9.3
MineBench — Bradley-Terry RatingGames18461992+8.8
Legal Research Bench CriminalLegal36.828.1−8.8
WCB (SecureBio)Science59.751.5−8.2
LiveBench 2.0Agents68.776.7+8.0
Humanity's Last ExamGeneral QA43.851.1+7.3
FrontierMath Tiers 1-3 (v2) - PrivateMath85.378.2−7.0
Legal Research Bench HealthLegal66.773.3+6.7
FrontierMath Tier 4 (v2) - PrivateMath72.565.9−6.6
APEX-Accounting 1.1Finance & Accounting56.362.4+6.1
LAB-Bench CloningScenarios (SecureBio)Science97.391.5−5.8
APEX-Agents 1.1Professional Work70.375.9+5.6
GPQA-Bio (SecureBio)Science87.482.2−5.2
CharXiv ReasoningMultimodal84.189.0+4.9
GBENCH One-shot Coding GScoreGames62.258.2−4.0
NoCoT-Bench NCKI (kspine v3)General QA111116+3.9
Creative Writing v3 Rubric ScoreChat & Writing15.916.6+3.7
MBCT (SecureBio)Science57.160.7+3.6
Tax Agent BenchFinance & Accounting60.563.9+3.4
LMCAGeneral QA54.357.6+3.3
LAB-Bench LitQA2 (SecureBio)Science49.853.1+3.3
HPCT (SecureBio)Science66.863.5−3.3
Legal Research Bench Administrative/RegulatoryLegal53.056.1+3.0
ACCoRDGeneral QA80.577.5−3.0
MMMU ProMultimodal81.284.1+2.9
VCT (SecureBio)Science52.349.4−2.9
DTBenchGeneral QA96.093.3−2.7
Legal Research Bench Constitutional/Civil RightsLegal28.931.6+2.6
Bio Capabilities Index (SecureBio)Indexes158161+2.6
DAXBench v2Code88.386.2−2.1
Longform Creative WritingChat & Writing78.279.9+1.7
Epoch Capabilities Index (ECI)Indexes160162+1.6
NoCoT-Bench NCRI 15.2General QA100103+1.4
VCT v2 (SecureBio)Science45.346.2+0.9
ALE-BenchCode21082132+0.8
SimpleQA Verified (Epoch anti-abstention)General QA63.062.2−0.8
Conceptual Reasoning IndexIndexes67.367.8+0.7
PubMedQA (SecureBio)Health & Medicine80.581.2+0.7
LMArena Text Arena EloChat & Writing14811483+0.3
LAB-Bench ProtocolQA (SecureBio)Science77.978.1+0.2
Legal Research Bench FamilyLegal13.613.60.0
OTIS Mock AIME 2024-2025Math100.0100.00.0

GPT-5.5vsKimi K3.1 Pronormalised score marginsData Nov 15, 2026Data: Vals AI, SecureBio, Epoch AI and 21 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…