Atlas

Compare Models

Opus 4.8 vs Kimi K3.1 Pro

Kimi K3.1 Pro leads by 4.1 capability points (≈95% interval +2.0 to +6.2): a clear lead over Opus 4.8 · P(higher) ≈ 100%

Wins 55 of 74 shared benchmarks (19 lost)

Capability vs. relative cost
Opus 4.8Kimi 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.

Opus 4.8vsKimi K3.1 Procapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins19 – 55 · 74 shared

Comparing matching setups only (25 benchmarks excluded) · suites counted once
← Opus 4.8Kimi K3.1 Pro →

Opus 4.8vsKimi K3.1 Prowins by categoryData Nov 15, 2026Data: SecureBio, Vals AI, Moonshot AI and 23 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 Opus 4.8.

ABC-Bench fragment design (SecureBio)Science19.884.2+64.4
Terminal-Bench 4.0Code23.665.7+42.1
Concept Synth INDUCTION Challenge64 Round 1General QA6.346.9+40.6
DeepsecBench — Revalidated V1 (233 issues)Code6.937.5+30.6
ZeroBench (Main questions, pass@5, with tools)Multimodal34.063.0+29.0
AutomationBench-AA 1.0.6Agents45.673.3+27.7
Terminal-Bench Science 0.1 (Vals)Science4.331.4+27.1
Legal Research Bench Civil LitigationLegal51.929.6−22.2
ZeroBench (Main questions, pass@5)Multimodal17.039.0+22.0
GameBench 2Games62.584.0+21.6
DeepSWE v1.1Code54.475.5+21.1
PerceptionBenchMultimodal47.268.3+21.1
Bug Hunt BenchCode14.334.3+20.0
Legal Research Bench HealthLegal53.373.3+20.0
AA-Briefcase v1.1 EloAgents13201678+19.6
Legal Research Bench CriminalLegal45.628.1−17.5
LMArena Code Arena WebDevCode15551688+17.3
BuseyBench SVGMultimodal5.26.9+17.3
GDPval-AA v2.1Professional Work14611846+17.0
Design Arena Overall Code CategoriesCode12521353+16.7
Humanity's Last Exam DiamondGeneral QA23.139.4+16.3
Chartography (with tools)Multimodal75.090.2+15.2
WCB (SecureBio)Science66.751.5−15.2
Legal Research Bench Weighted PassLegal86.171.4−14.6
BabyVision with PythonMultimodal81.295.4+14.2
Legal Research Bench Business/CommercialLegal41.927.9−14.0
EnterpriseBench: CoreCraftProfessional Work52.366.1+13.8
Furniture Assembly BenchmarkMultimodal42.555.8+13.3
MineBench — Bradley-Terry RatingGames17861992+12.5
Legal Research Bench Administrative/RegulatoryLegal43.956.1+12.1
APEX-Agents 1.1Professional Work64.475.9+11.5
FrontierMath Tier 4 (v2) - PrivateMath56.165.9+9.8
SimpleQA Verified (Epoch anti-abstention)General QA53.062.2+9.2
Legal Research Bench ImmigrationLegal27.318.2−9.1
CharXiv ReasoningMultimodal80.589.0+8.5
MMMU ProMultimodal75.683.9+8.3
Legal Research Bench Constitutional/Civil RightsLegal39.531.6−7.9
Chess Puzzles (task version 2.0)Games52.045.0−7.0
Mystery Game PuzzlesGames36.042.0+6.0
Kimi Code Bench v2Code71.777.1+5.4
PubMedQA (SecureBio)Health & Medicine75.981.2+5.3
FrontierCode 2.0 - Pass RateCode17.122.0+4.8
FrontierCode 2.0 - ScoreCode15.319.9+4.6
Legal Research Bench FamilyLegal18.213.6−4.5
LAB-Bench CloningScenarios (SecureBio)Science95.991.5−4.4
LAB-Bench LitQA2 (SecureBio)Science48.753.1+4.4
MBCT (SecureBio)Science56.460.7+4.3
GBENCH One-shot Coding GScoreGames54.058.2+4.2
BrowseCompSearch88.592.1+3.6
OfficeQA ProProfessional Work63.967.4+3.5
WMDP-Bio (SecureBio)Science92.389.7−2.6
Vibe Code Bench v1.1Code82.784.9+2.2
FrontierMath Tiers 1-3 (v2) - PrivateMath80.078.2−1.8
Epoch Capabilities Index (ECI)Indexes160162+1.7
ABC-Bench liquid handling (SecureBio)Science100.098.3−1.7
Bio Capabilities Index (SecureBio)Indexes164161−1.7
OTIS Mock AIME 2024-2025Math98.3100.0+1.7
DTBenchGeneral QA94.993.3−1.6
Humanity's Last ExamGeneral QA49.851.1+1.3
GPQA-Bio (SecureBio)Science80.982.2+1.3
VCT (SecureBio)Science48.249.4+1.2
LAB-Bench ProtocolQA (SecureBio)Science76.978.1+1.2
ALE-BenchCode21022132+1.1
Tax Agent BenchFinance & Accounting64.963.9−1.0
Conceptual Reasoning IndexIndexes68.367.8−0.6
NoCoT-Bench NCKI (kspine v3)General QA117116−0.6
NoCoT-Bench NCRI 15.2General QA102103+0.5
HPCT (SecureBio)Science63.163.5+0.4
VCT v2 (SecureBio)Science45.946.2+0.3
MMLU-Bio (SecureBio)Science90.991.2+0.3
LMArena Text Arena EloChat & Writing14821483+0.2
ACCoRDGeneral QA77.477.5+0.1
LMCAGeneral QA57.557.6+0.1
LiveBench 2.0Agents76.876.70.0

Opus 4.8vsKimi K3.1 Pronormalised score marginsData Nov 15, 2026Data: SecureBio, Vals AI, Moonshot AI and 23 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…