Enter opens the highlighted result. Esc closes.

Kimi K2.5 vs Kimi K2.6

Updated October 21, 2026. 84 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.

Replace Kimi K2.5 Replace Kimi K2.6
Kimi K2.5Moonshot AI
Coding
58.5
Overall
23.5
Output $/M
$2.50
6
Kimi K2.6Moonshot AI
Coding
44.2
Overall
27.0
Output $/M
$3.41
76
vs boards won of 84 shared, 2 tied

Decision reading

Pick Kimi K2.5 for coding: it scores 58.5 on the Vibe Coding Index against 44.2 for Kimi K2.6.

Pick Kimi K2.6 when what matters is LMArena WebDev or AA Coding Index, where it leads.

Model AKimi K2.5Moonshot AI58.5/1002 independent benchmarks$0.24 per solved taskModel BKimi K2.6Moonshot AI44.2/1001 independent benchmark$0.35 per solved task
Shared results
84
Kimi K2.5 only
14
Kimi K2.6 only
25
Like-for-like categories
8 of 9

Shared results can support a head-to-head reading. Results only one model has describe coverage, not superiority.

Which one for your work

Each card is decided by one number both models carry. Where only one has it, the card says so and makes no call.

At a glance

Coding index
Kimi K2.5 58.5 from 2 benchmarks, Kimi K2.6 44.2 from 1 benchmark.
Widest category
Agentic work, 11.0 points apart on average over 10 shared benchmarks.
Widest single gap
Vending-Bench 2: Kimi K2.5 $1,198 against Kimi K2.6 $6,205, 31.3 points apart on a 0 to 100 scale.
Output price
Kimi K2.5 $2.50 against Kimi K2.6 $3.41 per million output tokens, 1.4x apart.
Context window
Both take 262K tokens.

Benchmark by benchmark

First the score: boards each model leads, per category, where both have a result. Then every board, a category to a fold; bars are each result on a 0 to 100 scale, so compare the two in a row, not one row with another.

Boards won, where both have a result
CategoryKimi K2.5BarsKimi K2.6Ties
Coding015
Intelligence02
Agentic work172 ties
Reasoning09
Math02
Knowledge29
Web design17
General225
All shared boards6762 ties
CodingLike-for-likeFour of these boards feed the Vibe Coding Index015
Coding benchmarks for Kimi K2.5 and Kimi K2.6
BenchmarkKimi K2.5Kimi K2.6Leader
SWE-bench Verified70.8%Not measuredOnly Kimi K2.5 measured
SWE-bench Multilingual67.3%Not measuredOnly Kimi K2.5 measured
LMArena WebDev14371509Kimi K2.6
AA Coding Index46.861.8Kimi K2.6
LiveCodeBench (Vals AI)83.9%86.8%Kimi K2.6
SWE-bench Verified (Vals AI)70.0%76.2%Kimi K2.6
Vibe Code Bench (Vals AI)17.5%37.9%Kimi K2.6
Code migration (Vals AI)7.0%27.8%Kimi K2.6
ProgramBench (Vals AI)Not measured0.0%Only Kimi K2.6 measured
Terminal-Bench 2.1 (Vals AI)42.0%53.6%Kimi K2.6
SciCodeNot measured51.5%Only Kimi K2.6 measured
Terminal-Bench Hard (Artificial Analysis run)34.8%43.9%Kimi K2.6
Terminal-Bench43.2%Not measuredOnly Kimi K2.5 measured
WeirdML45.6%55.9%Kimi K2.6
ALE-Bench821.61092.7Kimi K2.6
LMArena Text: coding14831489Kimi K2.6
LMArena Text: software and IT14711484Kimi K2.6
LMArena WebDev: React14291505Kimi K2.6
LMArena WebDev: HTML14261485Kimi K2.6
LMArena WebDev: from an image14391521Kimi K2.6
Neither model has a result on the other 18 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index02
Intelligence benchmarks for Kimi K2.5 and Kimi K2.6
BenchmarkKimi K2.5Kimi K2.6Leader
AA Intelligence Index23.527.0Kimi K2.6
Epoch Capabilities Index148.1150.9Kimi K2.6
Agentic workLike-for-likePublic boards, shown as published, kept out of the index17
Agentic work benchmarks for Kimi K2.5 and Kimi K2.6
BenchmarkKimi K2.5Kimi K2.6Leader
AA Agentic IndexNot measured20.5Only Kimi K2.6 measured
Design Arena: agents building web apps11361268Kimi K2.6
Design Arena: agents building full-stack apps11141155Kimi K2.6
Design Arena: agents building mobile apps11471156Kimi K2.6
Design Arena: agents building Android apps11241166Kimi K2.6
Design Arena: agents building Godot games12191120Kimi K2.5
Design Arena: agents building gamesNot measured1141Only Kimi K2.6 measured
Design Arena: HTML slidesNot measured1184Only Kimi K2.6 measured
Design Arena: PowerPoint slidesNot measured1128Only Kimi K2.6 measured
Design Arena: spreadsheetsNot measured1276Only Kimi K2.6 measured
Finance Agent (Vals AI)35.8%44.9%Kimi K2.6
Tax Agent Bench (Vals AI)Not measured12.4%Only Kimi K2.6 measured
Legal Research Bench (Vals AI)15.9%15.9%Tie
Excel Modeling Benchmark (Vals AI)28.5%57.9%Kimi K2.6
τ²-Bench Telecom95.9%95.9%Tie
OSWorld 2.0Not measured4.6%Only Kimi K2.6 measured
OSWorld63.3%Not measuredOnly Kimi K2.5 measured
Vending-Bench 2$1,198$6,205Kimi K2.6
Neither model has a result on the other 21 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index09
Reasoning benchmarks for Kimi K2.5 and Kimi K2.6
BenchmarkKimi K2.5Kimi K2.6Leader
GPQA Diamond84.1%89.1%Kimi K2.6
MMMU Pro (Vals AI)84.3%86.3%Kimi K2.6
GPQA Diamond (Artificial Analysis run)87.9%91.1%Kimi K2.6
Humanity's Last Exam (Artificial Analysis run)30.7%37.5%Kimi K2.6
GPQA Diamond (Epoch AI run)Not measured90.8%Only Kimi K2.6 measured
ARC-AGI65.3%Not measuredOnly Kimi K2.5 measured
ARC-AGI-211.8%Not measuredOnly Kimi K2.5 measured
SimpleBench46.8%Not measuredOnly Kimi K2.5 measured
Chess puzzlesNot measured20.0%Only Kimi K2.6 measured
Mystery game puzzlesNot measured18.0%Only Kimi K2.6 measured
EBR-benchNot measured2.4%Only Kimi K2.6 measured
LMArena Text: hard prompts14531470Kimi K2.6
LMArena Text: expert14661491Kimi K2.6
LMArena Text: hard prompts in English14601471Kimi K2.6
LMArena Vision: diagrams12821298Kimi K2.6
LMArena Vision: homework12971310Kimi K2.6
Neither model has a result on the other 5 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index02
Math benchmarks for Kimi K2.5 and Kimi K2.6
BenchmarkKimi K2.5Kimi K2.6Leader
AIME95.6%Not measuredOnly Kimi K2.5 measured
OTIS Mock AIMENot measured96.1%Only Kimi K2.6 measured
FrontierMath, tiers 1 to 3Not measured57.2%Only Kimi K2.6 measured
FrontierMath, tier 4Not measured25.6%Only Kimi K2.6 measured
ProofBenchNot measured16.0%Only Kimi K2.6 measured
LMArena Text: math14701474Kimi K2.6
LMArena Text: mathematical work14731480Kimi K2.6
Neither model has a result on the other 6 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index29
Knowledge benchmarks for Kimi K2.5 and Kimi K2.6
BenchmarkKimi K2.5Kimi K2.6Leader
MMLU Pro85.9%87.6%Kimi K2.6
LegalBench (Vals AI)Not measured84.7%Only Kimi K2.6 measured
CaseLaw (Vals AI)58.7%61.2%Kimi K2.6
CorpFin (Vals AI)68.3%66.7%Kimi K2.5
TaxEval (Vals AI)74.2%74.7%Kimi K2.6
MortgageTax (Vals AI)66.5%65.8%Kimi K2.5
MedQA (Vals AI)94.4%Not measuredOnly Kimi K2.5 measured
MedCode (Vals AI)39.3%40.1%Kimi K2.6
SimpleQA Verified34.3%34.9%Kimi K2.6
LMArena Text: science14591468Kimi K2.6
LMArena Text: medicine and healthcare14481457Kimi K2.6
LMArena Text: legal and government14411462Kimi K2.6
LMArena Text: business and finance14271445Kimi K2.6
Neither model has a result on the other 4 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index17
Web design benchmarks for Kimi K2.5 and Kimi K2.6
BenchmarkKimi K2.5Kimi K2.6Leader
Design Arena: websites12491269Kimi K2.6
Design Arena: UI components12361266Kimi K2.6
Design Arena: game dev12071251Kimi K2.6
Design Arena: data viz12191249Kimi K2.6
Design Arena: 3D12051268Kimi K2.6
Design Arena: all code categories12361272Kimi K2.6
Design Arena: SVG11491181Kimi K2.6
Design Arena: ASCII art11651164Kimi K2.5
Neither model has a result on the other 1 board here
GeneralLike-for-likePublic boards, shown as published, kept out of the index225
General benchmarks for Kimi K2.5 and Kimi K2.6
BenchmarkKimi K2.5Kimi K2.6Leader
Public Benefits Bench (Vals AI)Not measured56.6%Only Kimi K2.6 measured
SAGE (Vals AI)49.9%50.2%Kimi K2.6
MedScribe (Vals AI)76.4%78.2%Kimi K2.6
Vals Multimodal IndexNot measured56.4%Only Kimi K2.6 measured
IFBench70.2%76.0%Kimi K2.6
Long context reasoning (AA-LCR)78.0%81.0%Kimi K2.6
DTBenchNot measured90.9%Only Kimi K2.6 measured
LMCANot measured37.3%Only Kimi K2.6 measured
Fiction.LiveBench86.1%Not measuredOnly Kimi K2.5 measured
CL-bench19.3%Not measuredOnly Kimi K2.5 measured
Furniture assemblyNot measured21.7%Only Kimi K2.6 measured
ExploitBenchNot measured18.4%Only Kimi K2.6 measured
CL-bench Life13.2%Not measuredOnly Kimi K2.5 measured
LMArena Text14451455Kimi K2.6
LMArena Text: instruction following14311451Kimi K2.6
LMArena Text: creative writing14221434Kimi K2.6
LMArena Vision12691282Kimi K2.6
LMArena Text: English14531460Kimi K2.6
LMArena Text: non-English14331446Kimi K2.6
LMArena Text: Chinese14941520Kimi K2.6
LMArena Text: German14411451Kimi K2.6
LMArena Text: Japanese14231443Kimi K2.6
LMArena Text: Korean14121427Kimi K2.6
LMArena Text: French14541471Kimi K2.6
LMArena Text: Spanish14501464Kimi K2.6
LMArena Text: Russian14351446Kimi K2.6
LMArena Text: Polish14451438Kimi K2.5
LMArena Text: multi-turn14441453Kimi K2.6
LMArena Text: longer queries14451468Kimi K2.6
LMArena Text: writing and language14251439Kimi K2.6
LMArena Text: entertainment and media14171428Kimi K2.6
LMArena Vision: English12741287Kimi K2.6
LMArena Vision: Chinese13281346Kimi K2.6
LMArena Vision: reading text12761291Kimi K2.6
LMArena Vision: humor12841280Kimi K2.5
LMArena Vision: captioning1257Not measuredOnly Kimi K2.5 measured
LMArena Vision: recognizing things1285Not measuredOnly Kimi K2.5 measured
LMArena Vision: creative writing12691273Kimi K2.6
LMArena DocumentNot measured1451Only Kimi K2.6 measured
Neither model has a result on the other 7 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for Kimi K2.5 and Kimi K2.6
BenchmarkKimi K2.5Kimi K2.6Leader
Neither model has a result on 5 boards here

What each workload costs

Three fixed token mixes at each model's listed price, so per-token rates become a bill.

Chat turn

1K input + 500 output tokens

Kimi K2.5$0.0018Fits in one request
Kimi K2.6$0.0024Fits in one request

Kimi K2.5 costs less

Repository review

50K input + 3K output tokens

Kimi K2.5$0.033Fits in one request
Kimi K2.6$0.043Fits in one request

Kimi K2.5 costs less

Agent loop

200K input + 10K output tokens

Kimi K2.5$0.125Fits in one request
Kimi K2.6$0.164Fits in one request

Kimi K2.5 costs less

Listed OpenRouter rates, without cache or batch discounts. A workload that does not fit a model's context window would need splitting.

Specification differences

The leading value in a row is marked; a row with a missing value names no leader. Our two rows are marked in red.

Specifications of Kimi K2.5 and Kimi K2.6
SpecificationMoonshotAI Kimi K2.5MoonshotAI Kimi K2.6
MakerMoonshot AIMoonshot AI
WeightsOpen weightsOpen weights
ReleasedJanuary 27, 2026April 20, 2026
Context window262K262K
Input $/Mtok$0.50$0.65
Output $/Mtok$2.50$3.41
Coding index58.544.2
Independent measurements behind it21
Cost per solved task$0.24$0.35
SourceOpenRouter listingOpenRouter listing

Questions

Is Kimi K2.5 better than Kimi K2.6 for coding?

Pick Kimi K2.5 for coding: it scores 58.5 on the Vibe Coding Index against 44.2 for Kimi K2.6. Pick Kimi K2.6 when what matters is LMArena WebDev or AA Coding Index, where it leads.

Which is cheaper, Kimi K2.5 or Kimi K2.6?

Kimi K2.5 costs $0.50 per million input tokens and $2.50 per million output tokens; Kimi K2.6 costs $0.65 and $3.41. Per solved SWE-bench Verified task, Kimi K2.5 costs $0.24 and Kimi K2.6 $0.35.

Which has the bigger context window, Kimi K2.5 or Kimi K2.6?

Neither: both take 262K tokens.

How many benchmarks is the Kimi K2.5 vs Kimi K2.6 comparison based on?

84 benchmarks have a result for both models as of October 21, 2026. Kimi K2.5's coding index rests on 2 independent benchmarks and Kimi K2.6's on 1. Where only one model has a result, the page says so instead of naming a winner.

Compare next