Kimi K3.1 Pro

Kimi K3.1 Pro ranks first on 9 of the 149 benchmarks with five or more models scored.

Moonshot AI Released October 23rd, 2026 Compare with others

156

Benchmarks with a score

Ranked first, out of five or more models

In the top three

Moonshot AI Oct 2026

Kimi K3.1 Pro

Open weights

161.4 #16 of 256

Epoch Capabilities Index

Self-Reported Index 170.0 #4 of 404

MMMU-Pro 84.1% #2 /90
Terminal-Bench 4.0 60.9% #3 /40
GPQA Diamond 94.2% #12 /229

Where it ranks

Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.

Knowledge and language

Benchmark	Score	Rank	Source
Creative Writing v3	83.2	#1 of 120	Official leaderboard
Humanity's Last Exam	70%	#2 of 154	Lab model card
LMArena (Elo)	1,483	#21 of 335	Official leaderboard
DesignArena ASCII Art	1,302	#8 of 77	Official leaderboard
Longform Creative Writing (EQ-Bench)	79.9	#16 of 119	Official leaderboard
Lech Mazur Creative Writing	2.72	#7 of 50	Official leaderboard
Extended NYT Connections	93.4%	#15 of 103	Official leaderboard
SimpleQA Verified	62.2%	#17 of 94	Epoch AI run
BullshitBench V2	60%	#36 of 129	Official leaderboard
Kimi Hallu Bench	5.40%	#2 of 6	Official leaderboard
HLE-Diamond	38.7%	#13 of 20	Official leaderboard
Design Bench (Moonshot)	957	#1 of 1	Lab model card
HANDBOOK.md	44.6%	#1 of 1	Lab model card

Math and science

Benchmark	Score	Rank	Source
GPQA Diamond	94.2%	#12 of 229	Epoch AI run
OTIS Mock AIME 2024-2025	100%	#13 of 212	Epoch AI run
Prophet Arena	86.2	#3 of 24	Official leaderboard
DTBench	93.3%	#34 of 180	Official leaderboard
ProofBench v1.1	91%	#11 of 52	Official leaderboard
FrontierMath Tiers 1-3 v2	78.3%	#23 of 92	Epoch AI run
FrontierMath Tier 4 v2	65.8%	#19 of 74	Epoch AI run
BioMysteryBench	75.6%	#8 of 30	Official leaderboard
Science's Last Exam	28.9	#4 of 9	Official leaderboard
FrontierPhysics	6.50%	#10 of 22	Official leaderboard
Riemann-bench	60.8%	#10 of 20	Official leaderboard
BrokenArXiv	61.9%	#3 of 6	Official leaderboard
MysteryMechanism	26.1%	#17 of 31	Official leaderboard
ArXivMath	58.3%	#4 of 6	Official leaderboard
ArXivLean	17%	#5 of 7	Official leaderboard
Terminal-Bench-Science	38.1%	#9 of 12	Lab model card

Coding

Benchmark	Score	Rank	Source
DesignArena UI Component	1,388	#2 of 139	Official leaderboard
DesignArena SVG	1,327	#3 of 103	Official leaderboard
DesignArena 3D Design	1,446	#5 of 136	Official leaderboard
DesignArena Game Dev	1,406	#6 of 143	Official leaderboard
OpenDesign Arena	91.0	#1 of 21	Official leaderboard
DesignArena Website	1,323	#9 of 151	Official leaderboard
DesignArena Full Stack	1,328	#3 of 49	Official leaderboard
DesignArena Code	1,351	#9 of 143	Official leaderboard
DesignArena React Native	1,273	#4 of 56	Official leaderboard
DesignArena Frontend Web App	1,296	#4 of 53	Official leaderboard
Code Arena	1,688	#10 of 128	Official leaderboard
DeepSWE	75.5%	#5 of 56	Lab model card
DesignArena Agentic Overall	1,283	#8 of 74	Official leaderboard
ALE-Bench	2,132	#16 of 126	Official leaderboard
DesignArena Data Visualization	1,304	#15 of 118	Official leaderboard
Tencent WorkBuddy Bench	78.7%	#2 of 14	Official leaderboard
DesignArena Image to HTML	1,254	#8 of 51	Official leaderboard
Vibe Code Bench	84.9%	#20 of 120	Official leaderboard
IOI v2 (Vals)	81.9%	#11 of 45	Official leaderboard
Kimi Code Bench 3.0 Long-horizon	61.2%	#2 of 6	Official leaderboard
Kimi Code Bench v2	77.1%	#1 of 3	Lab model card
PostTrainBench 1.2	36.5%	#1 of 3	Lab model card
DAXBench v2	83.9%	#16 of 47	Official leaderboard
Android Bench 2.0	24%	#6 of 17	Official leaderboard
SWE-Marathon	51.7%	#4 of 11	Lab model card
ProgramBench almost solved	31.5%	#3 of 8	Lab model card
Real-SWE 1.1	39.2%	#7 of 15	Official leaderboard
SWE Atlas v1.1	61.1%	#14 of 28	Official leaderboard
sudo L7	16.7%	#9 of 18	Official leaderboard
WeirdML v3	9.89%	#15 of 27	Official leaderboard
OWE-Bench	13.6%	#9 of 14	Official leaderboard
DAYJOB: Data Science	4.70%	#10 of 15	Official leaderboard
DesignArena Android (Kotlin)	1,149	#31 of 46	Official leaderboard
FrontierCode 2.0	19.9%	#23 of 32	Official leaderboard
Brokk Power Ranking	3.50	#16 of 19	Official leaderboard

Agents and tools

Benchmark	Score	Rank	Source
AutomationBench	73.3%	#1 of 47	Lab model card
BrowseComp	92.1%	#3 of 81	Lab model card
Terminal-Bench 4.0	60.9%	#3 of 40	Official leaderboard
RuneBench	6.48	#7 of 76	Official leaderboard
Terminal-Bench 4.1	59.7%	#6 of 43	Official leaderboard
DuelLab GameBench 2	84.0	#10 of 57	Official leaderboard
Mystery Game Puzzles	42%	#20 of 84	Epoch AI run
Agent Arena	4.89%	#15 of 56	Official leaderboard
BU Bench V2.1	76.7	#5 of 16	Official leaderboard
Chess Puzzles	45%	#32 of 93	Epoch AI run
Collaborative Intelligence Benchmark	59.9%	#7 of 18	Official leaderboard
OSWorld 2.1	79.6%	#3 of 7	Lab model card

Professional work

Benchmark	Score	Rank	Source
Harvey LAB (Legal Agent Benchmark)	23.7%	#1 of 14	Lab model card
PharmaBench (micro1)	27.5%	#1 of 13	Official leaderboard
APEX-Agents	67.2%	#1 of 12	Lab model card
YC-Bench	$2,386,003	#7 of 59	Official leaderboard
DesignArena Python-PPTX Slides	1,290	#5 of 36	Official leaderboard
EnterpriseBench: CoreCraft	66.1%	#8 of 51	Official leaderboard
BenchCAD	92.7%	#2 of 11	Lab model card
APEX-1	69.3%	#14 of 71	Official leaderboard
GDPval-AA v2.1	1,846	#2 of 10	Lab model card
Agents' Last Exam v1.1	43.4%	#3 of 14	Official leaderboard
APEX-Agents 1.1	75.9%	#15 of 65	Official leaderboard
FrontierFinance	51.7%	#4 of 17	Official leaderboard
APEX-Accounting	62.4%	#14 of 49	Official leaderboard
Vals Index v2.2	62.5%	#11 of 37	Official leaderboard
Agents' Last Exam	43.2%	#8 of 26	Lab model card
Tuesday Work Index	60.2	#14 of 44	Official leaderboard
DesignArena PDF	1,228	#12 of 37	Official leaderboard
DAYJOB: Finance	17%	#13 of 39	Official leaderboard
Bedside Bench	91.5%	#6 of 17	Official leaderboard
OfficeQA Pro	67.4%	#6 of 16	Lab model card
Excel Modeling Benchmark v1.1	70%	#18 of 42	Official leaderboard
PRBench	47.9%	#21 of 48	Official leaderboard
DAYJOB: Law	23.9%	#8 of 16	Official leaderboard
AA-Briefcase v1.1	1,678	#4 of 8	Lab model card
Legal Research Bench	37.5%	#47 of 93	Official leaderboard
DeepFinance	34.6%	#12 of 23	Official leaderboard
Harvey LAB v1.1 (HLAB-Vals-1.1)	11.7%	#20 of 38	Official leaderboard
Tax Agent Bench	24.9%	#46 of 84	Official leaderboard
Vals Finance Agent v2.1	61.5%	#22 of 40	Official leaderboard
DAYJOB: Accounting	13.6%	#10 of 17	Official leaderboard
RWE-bench	1.00%	#9 of 15	Official leaderboard
Vals Finance Index	56%	#25 of 40	Official leaderboard
Vals Legal Index	67%	#30 of 38	Official leaderboard
LegalReasonBench	36.4%	#9 of 10	Official leaderboard
Business Operation (Moonshot)	37.8%	#1 of 1	Lab model card
DECK-Arena	1,456	#1 of 1	Lab model card
OfficeQA Pro V2	64.4%	#1 of 1	Lab model card

Vision, audio and robotics

Benchmark	Score	Rank	Source
MMMU-Pro	84.1%	#2 of 90	Lab model card
BabyVision	95.4%	#1 of 26	Lab model card
Chartography	90.2%	#1 of 14	Lab model card
MineBench	1,970	#10 of 75	Official leaderboard
VoxelBench	2,177	#8 of 56	Official leaderboard
CharXiv	89%	#11 of 75	Lab model card
ZeroBench	39%	#5 of 31	Lab model card
DesignArena Code to Video	1,232	#6 of 29	Official leaderboard
Furniture Assembly Benchmark	55.8%	#13 of 39	Epoch AI run
PerceptionBench (Moonshot)	68.3%	#2 of 5	Lab model card
Humanity's Sixth Sense	33.2%	#14 of 34	Official leaderboard
Frontier3DBench	56.0	#5 of 11	Official leaderboard
GDP.pdf	46.6%	#10 of 19	Official leaderboard
Embodied First Exam	10.1%	#5 of 9	Official leaderboard

Safety and security

Benchmark	Score	Rank	Source
ABC Bench Screening Evasion	82.9%	#1 of 17	Official leaderboard
Molecular Biology Capabilities Test	60.7%	#6 of 90	Official leaderboard
nocot-bench (NCRI)	102.8	#17 of 185	Official leaderboard
DrugDiscoveryBench	63.4%	#3 of 20	Official leaderboard
DeepsecBench	37.5	#5 of 29	Official leaderboard
BioTIER Permit	99.9%	#12 of 69	Official leaderboard
Human Pathogen Capabilities Test	63.5%	#16 of 86	Official leaderboard
SRE-Bench	27.1%	#8 of 35	Official leaderboard
ABC Bench Liquid-Handling Robot	98.3%	#10 of 34	Official leaderboard
World-Class Bio	51.5%	#25 of 74	Official leaderboard
Human Pathogen Capabilities Test v2 (HPCT v2)	61.7%	#25 of 71	Official leaderboard
VCT-v2	46.2%	#9 of 25	Official leaderboard
FORTRESS	18.6%	#29 of 73	Official leaderboard
ExploitBench	74.7%	#9 of 21	Lab model card
Biopharma Bench	55%	#6 of 13	Official leaderboard
CyberOffense	63.2%	#3 of 6	Official leaderboard
xdev-bench-hard	81%	#3 of 5	Official leaderboard
BioTIER Refuse	32.6%	#43 of 69	Official leaderboard
ABC Bench Fragment Design	84.2%	#18 of 27	Official leaderboard
ExploitGym	20.7%	#10 of 15	Lab model card
CWE-bench 2	45.3%	#13 of 14	Official leaderboard

Other

Benchmark	Score	Rank	Source
LMCA	57.6%	#13 of 147	Official leaderboard
Conceptual Reasoning Index	67.8	#14 of 151	Official leaderboard
ACCoRD	77.5	#20 of 152	Official leaderboard
SuperCLUE	61.5	#3 of 16	Official leaderboard
Political Science LLM Benchmark	68.6%	#8 of 42	Official leaderboard
CAIS Text Capabilities Index	46.4	#14 of 68	Official leaderboard
Vals Science Index	44.4%	#10 of 25	Official leaderboard
LiveBench	68.7%	#13 of 31	Official leaderboard

Its lab's numbers put it at 170.0, 8.7 points above its ECI

The Self-Reported Index fits one score to the 20 capability results Moonshot AI reported for this model, on the scale of the Epoch Capabilities Index. Labs choose which benchmarks to show and how to run them, so read it as the lab's claim, not an independent measurement. The 90% interval runs from 166.3 to 173.8. How we compute it.

Its lab's numbers next to 9 leaderboards

Where the benchmark's own leaderboard, or Epoch AI's run, also has this model. Gaps mostly come from different harnesses, effort levels and task subsets. Every pair.

Benchmark	Lab says	Board says	Gap
GDP.pdf	34.6% GDP.pdf	46.6% Leaderboard, max	−12 pts
FrontierFinance	58.6% Frontier Finance	51.7% Leaderboard, max, Finance Agent v2	+6.9 pts
BioMysteryBench	80.6% BioMysteryBench (v11)	75.6% Leaderboard, Vals AI harness, v11 full set	+5.0 pts
Terminal-Bench 4.0	65.7% Terminal-Bench 4.0 (Claude Code)	60.9% Leaderboard, max, Claude Code	+4.8 pts
Kimi Hallu Bench	8.30% Kimi Hallu Bench Hard w/ tools (Internal, lower is better)	5.40% Leaderboard, without tools	−2.9 pts
Business Operation (Moonshot)	37.8% Business Operation (Internal)	37% Leaderboard	+0.8 pts
GPQA Diamond	94.2% GPQA-Diamond	94.2% Epoch AI run, Together, max	Same
Kimi Code Bench 3.0 Long-horizon	61.2% Kimi Code Bench 3.0 Long-horizon (Internal)	61.2% Leaderboard, max	Same
xdev-bench-hard	81% xdev-bench-hard (Internal, pass@3)	81% Leaderboard, kimi-code	Same

Moonshot AI's models over time

Moonshot AI's models in color, against every other model on the Epoch Capabilities Index, by release date.

Source: Epoch Capabilities Index, Epoch AI (CC BY 4.0)

Its launch post

The benchmarks the lab chose to headline, with the scores it gave.

Kimi K3.1 Pro Tech Blog: Advancing Open Intelligence Oct 24, 2026

Terminal-Bench 4.0 (Claude Code) 65.7
Terminal-Bench-Science 0.1 (Claude Code) 38.1
SWE Marathon v1.1 (Claude Code) 51.7
BenchCAD w/ tools (w/o tools 72.8) 92.7
Program Bench (binary score, >=95% counts as solved) 31.5
DeepSWE v1.1 (Claude Code) 75.5
PostTrain Bench v1.2 (H20 GPUs) 36.5
Kimi Code Bench 2.0 (Internal) 77.1
Kimi Code Bench 3.0 Long-horizon (Internal) 61.2
Best of HLE-Full w/ tools 70.0 / HLE-Full 51.1 70
BrowseComp 92.1
BioMysteryBench (v11) 80.6
CharXiv (RQ) 89
MMMU-Pro 84.1
GDP.pdf 34.6
ChartoGraphy w/ python 90.2
BabyVision w/ python 95.4
ZeroBench_main (avg@5 / pass@5) 29.8 / 39.0 39
Agents Last Exam (ALL) (CLI 44.2) 43.2
Automation Bench-AA 73.3
GDPval-AA v2.1 (Elo-score) 1846
AA-Briefcase v1.1 (Elo-score) 1678
APEX-Agents 67.2
OfficeQA Pro v2 64.4
OfficeQA Pro 67.4
Frontier Finance 58.6
Harvey Legal Agent (public 1,251 tasks, Stirrup harness, all-pass rate) 23.7
HANDBOOK.md 44.6
OSWorld V2.1 79.6
Business Operation (Internal) 37.8
DECK-Arena (Internal, Elo-score) 1456
Design Bench (Internal, Elo-score) 957
Kimi Hallu Bench Hard w/ tools (Internal, lower is better) 8.3
GPQA-Diamond 94.2
PerceptionBench 68.3
ExploitBench (pass@5 max capability score, 300 turns) 74.7
ExploitGym (6-hour budget) 20.7
xdev-bench-hard (Internal, pass@3) 81