Kimi K2 Thinking

Kimi K2 Thinking places in the top three on 4 benchmarks.

Moonshot AI Released November 6th, 2025 Family Kimi K2 The lab's model card Compare with others

Benchmarks with a score

Ranked first, out of five or more models

In the top three

Moonshot AI Nov 2025

Kimi K2 Thinking

Open weights Flagship

145.4 #93 of 256

Epoch Capabilities Index

Self-Reported Index 146.8 #117 of 404

Terminal-Bench 2.0 35.7% #27 /43
SciCode 44.8% #19 /49
MultiChallenge 55.4% #25 /42

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
MMLU-Redux	94.4%	#5 of 45	Lab model card
MMLU-Pro	84.6%	#37 of 167	Lab model card
FRAMES	87%	#2 of 9	Lab model card
Seal-0	56.3%	#2 of 7	Lab model card
Creative Writing v3	73.3	#36 of 120	Official leaderboard
Longform Creative Writing (EQ-Bench)	69.0	#43 of 119	Official leaderboard
Humanity's Last Exam	44.9%	#56 of 154	Lab model card
CL-bench	17.6%	#14 of 24	Official leaderboard
MultiChallenge	55.4%	#25 of 42	Official leaderboard
SkateBench v2	43.9%	#16 of 21	Official leaderboard
Longform Writing	73.8%	#1 of 1	Lab model card

Math and science

Benchmark	Score	Rank	Source
SMT 2025	91%	#6 of 40	Official leaderboard
AIME 2025	94.5%	#24 of 143	Lab model card
CMIMC 2025	90.6%	#6 of 33	Official leaderboard
BRUMO 2025	93.3%	#15 of 42	Official leaderboard
GPQA Diamond	84.2%	#83 of 229	Epoch AI run
FrontierMath	21.4%	#23 of 55	Official leaderboard
OTIS Mock AIME 2024-2025	83.1%	#93 of 212	Epoch AI run
HMMT 2025	89.4%	#24 of 54	Lab model card
HiPhO	57.2%	#16 of 36	Official leaderboard
General365	53%	#13 of 23	Official leaderboard
HMMT Nov 2025	89.2%	#14 of 22	Official leaderboard
SimpleBench	39.6%	#76 of 106	Official leaderboard
IMO-AnswerBench	78.6%	#27 of 37	Lab model card
MathArena Apex Shortlist	47.3%	#31 of 38	Official leaderboard
SnorkelSequences	36%	#23 of 25	Official leaderboard
FrontierMath Tier 4	0%	#52 of 55	Epoch AI run
MathArena (Apex, IMO, USAMO)	0%	#47 of 47	Official leaderboard

Coding

Benchmark	Score	Rank	Source
OJBench	48.7%	#4 of 16	Lab model card
LiveCodeBench v6	83.1%	#25 of 90	Lab model card
METR Time Horizon	58 min	#10 of 31	Official leaderboard
SciCode	44.8%	#19 of 49	Lab model card
SWE-bench Verified	63.4%	#27 of 60	Official leaderboard
Multi-SWE-bench	41.9%	#6 of 10	Lab model card
WeirdML	42.8%	#68 of 112	Official leaderboard
Bird-SQL	59.9%	#8 of 13	Official leaderboard
Snorkel Agentic Coding	36.8%	#7 of 11	Official leaderboard
CF-Div2-Stepfun	67.9%	#6 of 8	Official leaderboard
ALE-Bench	1,103	#95 of 126	Official leaderboard
DesignArena Website	1,106	#115 of 151	Official leaderboard
OTelBench	7.25%	#11 of 14	Official leaderboard
WebDev Arena	1,337	#85 of 105	WebDev Arena Leaderboard, via Epoch AI
SWE-bench Multilingual	61.1%	#60 of 73	Lab model card
OpenHands Index	41%	#28 of 33	Official leaderboard
PostTrainBench	7.25%	#23 of 23	Official leaderboard

Agents and tools

Benchmark	Score	Rank	Source
Terminal-Bench	47.1%	#3 of 22	Lab model card
SnakeBench	29	#46 of 188	Official leaderboard
Step Race	3	#15 of 59	Official leaderboard
Elimination Game	4	#26 of 56	Official leaderboard
SnorkelWordle	85%	#10 of 19	Official leaderboard
Terminal-Bench 2.0	35.7%	#27 of 43	Official leaderboard
BrowseComp-ZH	62.3%	#12 of 19	Lab model card
BrowseComp	60.2%	#58 of 81	Lab model card
Pencil Puzzle Bench	0%	#39 of 53	Official leaderboard
BalatroBench	5.47	#12 of 16	Official leaderboard
AgencyBench	34.2%	#7 of 9	Official leaderboard
Agent Poker Bench	1,012	#14 of 17	Official leaderboard
VITA-Bench	12.8%	#15 of 18	Official leaderboard
Toolathlon	17.6%	#42 of 50	Official leaderboard
EnterpriseOps-Gym	22.4%	#11 of 12	Official leaderboard
ClawProBench	47.8%	#45 of 49	Official leaderboard
RWSearch	63%	#7 of 7	Official leaderboard

Professional work

Benchmark	Score	Rank	Source
CaseLaw v2	65.7%	#11 of 53	Official leaderboard
ProfBench	48.9%	#32 of 72	Official leaderboard
SnorkelUnderwrite	71.3%	#13 of 27	Official leaderboard
FinQA (Snorkel OpenEnv environment)	71.7%	#12 of 22	Official leaderboard
SnorkelFinance	71.7%	#12 of 22	Official leaderboard
HealthBench	58%	#17 of 31	Lab model card
TaxEval v2	71.7%	#71 of 129	Official leaderboard
CorpFin v2	60.6%	#67 of 119	Official leaderboard
Vals Finance Agent v1	26%	#28 of 47	Official leaderboard
LegalBench	80.2%	#85 of 138	Official leaderboard
PRBench	40.9%	#34 of 48	Official leaderboard
Physical-Engineering's Last Exam	0.06	#13 of 18	Official leaderboard
Finance Reasoning (Snorkel)	35%	#16 of 21	Official leaderboard
Vals Finance Agent	36.6%	#41 of 49	Official leaderboard
prinzbench	22.0	#22 of 26	Official leaderboard
APEX-Agents	4.00%	#8 of 8	Official leaderboard
FinSearchComp	47.4%	#7 of 7	Lab model card

Safety and security

Benchmark	Score	Rank	Source
BioTIER Permit	99.9%	#3 of 69	Official leaderboard
Lakera B3 (Backbone Breaker Benchmark)	65.7%	#13 of 35	Official leaderboard
Human Pathogen Capabilities Test	52.4%	#42 of 86	Official leaderboard
Human Pathogen Capabilities Test v2 (HPCT v2)	50.4%	#42 of 71	Official leaderboard
Molecular Biology Capabilities Test	46.1%	#56 of 90	Official leaderboard
World-Class Bio	38.9%	#49 of 74	Official leaderboard
Agentic BAIM-LLM Evaluation (ABLE)	48.1%	#12 of 18	Official leaderboard
BioTIER Refuse	16.3%	#63 of 69	Official leaderboard

Other

Benchmark	Score		Rank	Source
CAIS Text Capabilities Index	19.9		#51 of 68	Official leaderboard
CAIS Risk Index	63.9		#55 of 67	Official leaderboard

Its lab's numbers put it at 146.8, 1.4 points above its ECI

The Self-Reported Index fits one score to the 18 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 145.0 to 148.6. How we compute it.

Its lab's numbers next to 4 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
SWE-bench Verified	71.3% SWE-bench Verified, w/ tools	63.4% Leaderboard, mini-SWE-agent	+7.9 pts
HMMT 2025	89.4% HMMT25, no tools	93.3% Leaderboard	−3.9 pts
AIME 2025	94.5% AIME25, no tools	92.5% Leaderboard	+2.0 pts
GPQA Diamond	84.5% GPQA, no tools	84.2% Epoch AI run	+0.3 pts

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 K2 Thinking model card (Hugging Face) Nov 6, 2025