LLMBenchmark.ai

Enter opens the highlighted result. Esc closes.

Kimi K2.7 Code vs MiniMax M2.5

Updated October 21, 2026. 23 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.

Replace Kimi K2.7 Code Replace MiniMax M2.5
Kimi K2.7 CodeMoonshot AI
Coding
39.1
Overall
25.8
Output $/M
$3.35
21
MiniMax M2.5MiniMax
Coding
59.4
Overall
22.8
Output $/M
$0.95
2
vs boards won of 23 shared

Decision reading

Pick MiniMax M2.5 for coding: it scores 59.4 on the Vibe Coding Index against 39.1 for Kimi K2.7 Code.

Pick Kimi K2.7 Code when a solved task costs $0.35 against $0.62, or when the context window is larger, 262K against 205K tokens.

Model AKimi K2.7 CodeMoonshot AI39.1/1001 independent benchmark$0.35 per solved taskModel BMiniMax M2.5MiniMax59.4/1002 independent benchmarks$0.62 per solved task
Shared results
23
Kimi K2.7 Code only
28
MiniMax M2.5 only
44
Like-for-like categories
6 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 K2.7 Code 39.1 from 1 benchmark, MiniMax M2.5 59.4 from 2 benchmarks.
Widest category
Agentic work, 18.5 points apart on average over 2 shared benchmarks.
Widest single gap
Vibe Code Bench (Vals AI): Kimi K2.7 Code 47.2% against MiniMax M2.5 14.8%, 32.4 points apart on a 0 to 100 scale.
Output price
Kimi K2.7 Code $3.35 against MiniMax M2.5 $0.95 per million output tokens, 3.5x apart.
Context window
Kimi K2.7 Code takes 262K tokens, MiniMax M2.5 205K.

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 K2.7 CodeBarsMiniMax M2.5Ties
Coding80
Intelligence20
Agentic work11
Reasoning20
Web design70
General11
All shared boards212
CodingLike-for-likeFour of these boards feed the Vibe Coding Index80
Coding benchmarks for Kimi K2.7 Code and MiniMax M2.5
BenchmarkKimi K2.7 CodeMiniMax M2.5Leader
SWE-bench VerifiedNot measured75.8%Only MiniMax M2.5 measured
SWE-bench MultilingualNot measured68.3%Only MiniMax M2.5 measured
LMArena WebDev14741386Kimi K2.7 Code
AA Coding Index60.8Not measuredOnly Kimi K2.7 Code measured
LiveCodeBench (Vals AI)82.0%79.2%Kimi K2.7 Code
SWE-bench Verified (Vals AI)78.2%74.2%Kimi K2.7 Code
Vibe Code Bench (Vals AI)47.2%14.8%Kimi K2.7 Code
Code migration (Vals AI)25.4%Not measuredOnly Kimi K2.7 Code measured
ProgramBench (Vals AI)0.0%Not measuredOnly Kimi K2.7 Code measured
Terminal-Bench 2.1 (Vals AI)67.0%Not measuredOnly Kimi K2.7 Code measured
SkillsBench (Vals AI)50.0%Not measuredOnly Kimi K2.7 Code measured
SciCode47.8%Not measuredOnly Kimi K2.7 Code measured
Terminal-Bench Hard (Artificial Analysis run)44.7%34.8%Kimi K2.7 Code
Terminal-BenchNot measured42.7%Only MiniMax M2.5 measured
FrontierCode30.1%Not measuredOnly Kimi K2.7 Code measured
WeirdML54.1%Not measuredOnly Kimi K2.7 Code measured
ALE-Bench886.2618.2Kimi K2.7 Code
LMArena Text: codingNot measured1380Only MiniMax M2.5 measured
LMArena Text: software and ITNot measured1383Only MiniMax M2.5 measured
LMArena WebDev: React14751374Kimi K2.7 Code
LMArena WebDev: HTML14511414Kimi K2.7 Code
LMArena WebDev: from an image1513Not measuredOnly Kimi K2.7 Code measured
Neither model has a result on the other 16 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index20
Intelligence benchmarks for Kimi K2.7 Code and MiniMax M2.5
BenchmarkKimi K2.7 CodeMiniMax M2.5Leader
AA Intelligence Index25.822.8Kimi K2.7 Code
Epoch Capabilities Index150.1146.6Kimi K2.7 Code
Agentic workLike-for-likePublic boards, shown as published, kept out of the index11
Agentic work benchmarks for Kimi K2.7 Code and MiniMax M2.5
BenchmarkKimi K2.7 CodeMiniMax M2.5Leader
AA Agentic Index21.0Not measuredOnly Kimi K2.7 Code measured
Design Arena: agents building web apps1176Not measuredOnly Kimi K2.7 Code measured
Design Arena: agents building full-stack apps1175Not measuredOnly Kimi K2.7 Code measured
Design Arena: agents building mobile apps1143Not measuredOnly Kimi K2.7 Code measured
Design Arena: agents building Android apps1143Not measuredOnly Kimi K2.7 Code measured
Design Arena: agents building Godot games1147Not measuredOnly Kimi K2.7 Code measured
Design Arena: agents building games1130Not measuredOnly Kimi K2.7 Code measured
Design Arena: HTML slides1201Not measuredOnly Kimi K2.7 Code measured
Design Arena: PowerPoint slides1154Not measuredOnly Kimi K2.7 Code measured
τ²-Bench Telecom90.1%95.3%MiniMax M2.5
APEX-Agents37.6%Not measuredOnly Kimi K2.7 Code measured
Vending-Bench 2$5,083-$23Kimi K2.7 Code
Neither model has a result on the other 27 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index20
Reasoning benchmarks for Kimi K2.7 Code and MiniMax M2.5
BenchmarkKimi K2.7 CodeMiniMax M2.5Leader
GPQA DiamondNot measured82.1%Only MiniMax M2.5 measured
GPQA Diamond (Artificial Analysis run)89.6%84.8%Kimi K2.7 Code
Humanity's Last Exam (Artificial Analysis run)35.0%20.5%Kimi K2.7 Code
GPQA Diamond (Epoch AI run)87.9%Not measuredOnly Kimi K2.7 Code measured
ARC-AGINot measured63.7%Only MiniMax M2.5 measured
ARC-AGI-2Not measured4.9%Only MiniMax M2.5 measured
SimpleBench57.9%Not measuredOnly Kimi K2.7 Code measured
Chess puzzles14.0%Not measuredOnly Kimi K2.7 Code measured
LMArena Text: hard promptsNot measured1372Only MiniMax M2.5 measured
LMArena Text: expertNot measured1378Only MiniMax M2.5 measured
LMArena Text: hard prompts in EnglishNot measured1384Only MiniMax M2.5 measured
Neither model has a result on the other 10 boards here
MathDirectional onlyPublic boards, shown as published, kept out of the index7 boards, none shared
Math benchmarks for Kimi K2.7 Code and MiniMax M2.5
BenchmarkKimi K2.7 CodeMiniMax M2.5Leader
AIMENot measured88.8%Only MiniMax M2.5 measured
OTIS Mock AIME95.6%Not measuredOnly Kimi K2.7 Code measured
FrontierMath, tiers 1 to 354.0%Not measuredOnly Kimi K2.7 Code measured
FrontierMath, tier 412.2%Not measuredOnly Kimi K2.7 Code measured
ProofBenchNot measured4.0%Only MiniMax M2.5 measured
LMArena Text: mathNot measured1378Only MiniMax M2.5 measured
LMArena Text: mathematical workNot measured1378Only MiniMax M2.5 measured
Neither model has a result on the other 6 boards here
KnowledgeDirectional onlyPublic boards, shown as published, kept out of the index11 boards, none shared
Knowledge benchmarks for Kimi K2.7 Code and MiniMax M2.5
BenchmarkKimi K2.7 CodeMiniMax M2.5Leader
MMLU ProNot measured80.1%Only MiniMax M2.5 measured
LegalBench (Vals AI)Not measured80.0%Only MiniMax M2.5 measured
CaseLaw (Vals AI)Not measured53.5%Only MiniMax M2.5 measured
CorpFin (Vals AI)Not measured59.6%Only MiniMax M2.5 measured
TaxEval (Vals AI)Not measured68.2%Only MiniMax M2.5 measured
MedQA (Vals AI)Not measured92.5%Only MiniMax M2.5 measured
SimpleQA Verified36.5%Not measuredOnly Kimi K2.7 Code measured
LMArena Text: scienceNot measured1369Only MiniMax M2.5 measured
LMArena Text: medicine and healthcareNot measured1357Only MiniMax M2.5 measured
LMArena Text: legal and governmentNot measured1366Only MiniMax M2.5 measured
LMArena Text: business and financeNot measured1356Only MiniMax M2.5 measured
Neither model has a result on the other 6 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index70
Web design benchmarks for Kimi K2.7 Code and MiniMax M2.5
BenchmarkKimi K2.7 CodeMiniMax M2.5Leader
Design Arena: websites12641222Kimi K2.7 Code
Design Arena: UI components12591172Kimi K2.7 Code
Design Arena: game dev12201176Kimi K2.7 Code
Design Arena: data viz12261171Kimi K2.7 Code
Design Arena: 3D12411170Kimi K2.7 Code
Design Arena: all code categories12531207Kimi K2.7 Code
Design Arena: SVG11761147Kimi K2.7 Code
Design Arena: ASCII art1196Not measuredOnly Kimi K2.7 Code measured
Neither model has a result on the other 1 board here
GeneralLike-for-likePublic boards, shown as published, kept out of the index11
General benchmarks for Kimi K2.7 Code and MiniMax M2.5
BenchmarkKimi K2.7 CodeMiniMax M2.5Leader
IFBench63.1%71.6%MiniMax M2.5
Long context reasoning (AA-LCR)79.3%73.3%Kimi K2.7 Code
CL-benchNot measured11.4%Only MiniMax M2.5 measured
Surface Evolver Bench48.8%Not measuredOnly Kimi K2.7 Code measured
CL-bench LifeNot measured6.3%Only MiniMax M2.5 measured
LMArena TextNot measured1359Only MiniMax M2.5 measured
LMArena Text: instruction followingNot measured1353Only MiniMax M2.5 measured
LMArena Text: creative writingNot measured1331Only MiniMax M2.5 measured
LMArena Text: EnglishNot measured1380Only MiniMax M2.5 measured
LMArena Text: non-EnglishNot measured1338Only MiniMax M2.5 measured
LMArena Text: ChineseNot measured1393Only MiniMax M2.5 measured
LMArena Text: GermanNot measured1361Only MiniMax M2.5 measured
LMArena Text: JapaneseNot measured1171Only MiniMax M2.5 measured
LMArena Text: KoreanNot measured1233Only MiniMax M2.5 measured
LMArena Text: FrenchNot measured1363Only MiniMax M2.5 measured
LMArena Text: SpanishNot measured1353Only MiniMax M2.5 measured
LMArena Text: RussianNot measured1359Only MiniMax M2.5 measured
LMArena Text: PolishNot measured1334Only MiniMax M2.5 measured
LMArena Text: multi-turnNot measured1364Only MiniMax M2.5 measured
LMArena Text: longer queriesNot measured1366Only MiniMax M2.5 measured
LMArena Text: writing and languageNot measured1345Only MiniMax M2.5 measured
LMArena Text: entertainment and mediaNot measured1329Only MiniMax M2.5 measured
Neither model has a result on the other 24 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for Kimi K2.7 Code and MiniMax M2.5
BenchmarkKimi K2.7 CodeMiniMax M2.5Leader
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 K2.7 Code$0.0023Fits in one request
MiniMax M2.5$0.0007Fits in one request

MiniMax M2.5 costs less

Repository review

50K input + 3K output tokens

Kimi K2.7 Code$0.044Fits in one request
MiniMax M2.5$0.016Fits in one request

MiniMax M2.5 costs less

Agent loop

200K input + 10K output tokens

Kimi K2.7 Code$0.168Fits in one request
MiniMax M2.5$0.064Exceeds the context window

MiniMax M2.5 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 K2.7 Code and MiniMax M2.5
SpecificationMoonshotAI Kimi K2.7 CodeMiniMax M2.5
MakerMoonshot AIMiniMax
WeightsOpen weightsOpen weights
ReleasedJune 12, 2026February 12, 2026
Context window262K205K
Input $/Mtok$0.6712$0.27
Output $/Mtok$3.35$0.95
Coding index39.159.4
Independent measurements behind it12
Cost per solved task$0.35$0.62
SourceOpenRouter listingOpenRouter listing

Questions

Is Kimi K2.7 Code better than MiniMax M2.5 for coding?

Pick MiniMax M2.5 for coding: it scores 59.4 on the Vibe Coding Index against 39.1 for Kimi K2.7 Code. Pick Kimi K2.7 Code when a solved task costs $0.35 against $0.62, or when the context window is larger, 262K against 205K tokens.

Which is cheaper, Kimi K2.7 Code or MiniMax M2.5?

Kimi K2.7 Code costs $0.6712 per million input tokens and $3.35 per million output tokens; MiniMax M2.5 costs $0.27 and $0.95. Per solved SWE-bench Verified task, Kimi K2.7 Code costs $0.35 and MiniMax M2.5 $0.62.

Which has the bigger context window, Kimi K2.7 Code or MiniMax M2.5?

Kimi K2.7 Code, with 262K tokens against 205K.

How many benchmarks is the Kimi K2.7 Code vs MiniMax M2.5 comparison based on?

23 benchmarks have a result for both models as of October 21, 2026. Kimi K2.7 Code's coding index rests on 1 independent benchmark and MiniMax M2.5's on 2. Where only one model has a result, the page says so instead of naming a winner.

Compare next