Enter opens the highlighted result. Esc closes.

Kimi K2.5 vs MiniMax M3

MiniMax M3 leads Kimi K2.5 on coding, 39.2 to 38.6 on the Vibe Coding Index, and MiniMax M3 costs 2.4x less per output token ($1.20 vs $2.85 per million). Updated October 22, 2026; 82 benchmarks have a result for both.

Replace Kimi K2.5 Replace MiniMax M3
Kimi K2.5Moonshot AI
Coding
38.6
Overall
44.8
Output $/M
$2.85
44
MiniMax M3MiniMax
Coding
39.2
Overall
51.7
Output $/M
$1.20
38
vs boards won of 82 shared

Kimi K2.5 or MiniMax M3: which to pick

Kimi K2.5 and MiniMax M3 are level for coding, 38.6 against 39.2 on the Vibe Coding Index, so price and context decide.

Pick Kimi K2.5 for a lower cost per solved task ($0.24 against $0.56) and work measured by LiveCodeBench (Vals AI) or ALE-Bench, where it leads. Pick MiniMax M3 for output tokens 2.4x cheaper ($1.20 vs $2.85 per million) and a larger context window (1.05M against 262K tokens).

Model AKimi K2.5Moonshot AI38.6/1003 independent benchmarks$0.24 per solved taskModel BMiniMax M3MiniMax39.2/1006 independent benchmarks$0.56 per solved task
Shared results
82
Kimi K2.5 only
25
MiniMax M3 only
35
Like-for-like categories
8 of 9

Only benchmarks that tested both models count toward a pick; a result only one model has is shown but not compared.

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 38.6 from 3 benchmarks, MiniMax M3 39.2 from 6 benchmarks.
Widest category
Agentic work, 9.4 points apart on average over 9 shared benchmarks.
Widest single gap
Vibe Code Bench (Vals AI): Kimi K2.5 17.5% against MiniMax M3 47.6%, 30.0 points apart on a 0 to 100 scale.
Output price
Kimi K2.5 $2.85 against MiniMax M3 $1.20 per million output tokens, 2.4x apart.
Context window
Kimi K2.5 takes 262K tokens, MiniMax M3 1.05M.

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 K2.5BarsMiniMax M3Ties
Coding410
Intelligence11
Agentic work18
Reasoning73
Math20
Knowledge63
Web design17
General226
All shared boards4438
CodingLike-for-likeSix of these boards feed the Vibe Coding Index410
Coding benchmarks for Kimi K2.5 and MiniMax M3
BenchmarkKimi K2.5MiniMax M3Leader
SWE-bench Verified70.8%Not measuredOnly Kimi K2.5 measured
SWE-bench Multilingual67.3%Not measuredOnly Kimi K2.5 measured
LiveBench codingNot measured70.9%Only MiniMax M3 measured
LMArena WebDev14371482MiniMax M3
AA Coding Index46.858.6MiniMax M3
LiveCodeBench (Vals AI)83.9%82.2%Kimi K2.5
SWE-bench Verified (Vals AI)70.0%75.0%MiniMax M3
Terminal-Bench 4 (Vals AI)Not measured1.0%Only MiniMax M3 measured
Vibe Code Bench (Vals AI)17.5%47.6%MiniMax M3
Code migration (Vals AI)7.0%19.9%MiniMax M3
Terminal-Bench 2.1 (Vals AI)42.0%53.6%MiniMax M3
SWE-bench Multimodal (Vals AI)Not measured33.3%Only MiniMax M3 measured
Vibe Code Bench 1-100 (Vals AI)Not measured9.2%Only MiniMax M3 measured
SkillsBench (Vals AI)Not measured51.5%Only MiniMax M3 measured
CyberBench (Vals AI)Not measured65.0%Only MiniMax M3 measured
SciCodeNot measured47.1%Only MiniMax M3 measured
Terminal-Bench Hard (Artificial Analysis run)34.8%42.4%MiniMax M3
Terminal-Bench43.2%Not measuredOnly Kimi K2.5 measured
FrontierCodeNot measured14.7%Only MiniMax M3 measured
WeirdML45.6%Not measuredOnly Kimi K2.5 measured
ALE-Bench821.6640.0Kimi K2.5
LMArena Text: coding14831470Kimi K2.5
LMArena Text: software and IT14711466Kimi K2.5
LMArena WebDev: React14291478MiniMax M3
LMArena WebDev: HTML14261455MiniMax M3
LMArena WebDev: from an image14391470MiniMax M3
SWE Atlas: Codebase QnA13.1%Not measuredOnly Kimi K2.5 measured
SWE Atlas: Refactoring20.9%Not measuredOnly Kimi K2.5 measured
SWE Atlas: Test Writing25.8%Not measuredOnly Kimi K2.5 measured
Neither model has a result on the other 14 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index11
Intelligence benchmarks for Kimi K2.5 and MiniMax M3
BenchmarkKimi K2.5MiniMax M3Leader
AA Intelligence Index23.529.2MiniMax M3
Epoch Capabilities Index148.1146.9Kimi K2.5
Agentic workLike-for-likePublic boards, shown as published, kept out of the index18
Agentic work benchmarks for Kimi K2.5 and MiniMax M3
BenchmarkKimi K2.5MiniMax M3Leader
AA Agentic IndexNot measured29.5Only MiniMax M3 measured
Design Arena: agents building web apps11361200MiniMax M3
Design Arena: agents building full-stack apps11141197MiniMax M3
Design Arena: agents building mobile apps11471169MiniMax M3
Design Arena: agents building Android apps11181158MiniMax M3
Design Arena: agents building Godot games1219Not measuredOnly Kimi K2.5 measured
Design Arena: agents building gamesNot measured1159Only MiniMax M3 measured
Design Arena: HTML slidesNot measured1120Only MiniMax M3 measured
Design Arena: PowerPoint slidesNot measured1178Only MiniMax M3 measured
Design Arena: spreadsheetsNot measured1042Only MiniMax M3 measured
LiveBench researchNot measured68.0%Only MiniMax M3 measured
LiveBench browserNot measured8.0%Only MiniMax M3 measured
LiveBench long agenticNot measured12.4%Only MiniMax M3 measured
Finance Agent (Vals AI)35.8%48.3%MiniMax M3
Tax Agent Bench (Vals AI)Not measured20.6%Only MiniMax M3 measured
Legal Research Bench (Vals AI)15.9%29.8%MiniMax M3
Excel Modeling Benchmark (Vals AI)28.5%47.8%MiniMax M3
τ²-Bench Telecom95.9%88.9%Kimi K2.5
APEX-AgentsNot measured37.7%Only MiniMax M3 measured
OSWorld 2.0Not measured4.6%Only MiniMax M3 measured
OSWorld63.3%Not measuredOnly Kimi K2.5 measured
Vending-Bench 2$1,198$2,158MiniMax M3
MCP Atlas64.4%Not measuredOnly Kimi K2.5 measured
Neither model has a result on the other 19 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index73
Reasoning benchmarks for Kimi K2.5 and MiniMax M3
BenchmarkKimi K2.5MiniMax M3Leader
GPQA Diamond84.1%92.7%MiniMax M3
LiveBench reasoningNot measured74.5%Only MiniMax M3 measured
MMMU Pro (Vals AI)84.3%81.2%Kimi K2.5
GPQA Diamond (Artificial Analysis run)87.9%92.9%MiniMax M3
Humanity's Last Exam (Artificial Analysis run)30.7%39.0%MiniMax M3
GPQA Diamond (Epoch AI run)Not measured90.9%Only MiniMax M3 measured
ARC-AGI65.3%Not measuredOnly Kimi K2.5 measured
ARC-AGI-211.8%Not measuredOnly Kimi K2.5 measured
SimpleBench46.8%45.8%Kimi K2.5
Chess puzzlesNot measured13.0%Only MiniMax M3 measured
Mystery game puzzlesNot measured8.0%Only MiniMax M3 measured
LMArena Text: hard prompts14531447Kimi K2.5
LMArena Text: expert14661461Kimi K2.5
LMArena Text: hard prompts in English14601455Kimi K2.5
LMArena Vision: diagrams12821266Kimi K2.5
LMArena Vision: homework12971268Kimi K2.5
Kagi LLM Benchmark78.5%Not measuredOnly Kimi K2.5 measured
Neither model has a result on the other 6 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index20
Math benchmarks for Kimi K2.5 and MiniMax M3
BenchmarkKimi K2.5MiniMax M3Leader
AIME95.6%Not measuredOnly Kimi K2.5 measured
OTIS Mock AIMENot measured71.1%Only MiniMax M3 measured
ProofBenchNot measured18.0%Only MiniMax M3 measured
LMArena Text: math14701429Kimi K2.5
LMArena Text: mathematical work14731448Kimi K2.5
MathArena17.9%Not measuredOnly Kimi K2.5 measured
Neither model has a result on the other 8 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index63
Knowledge benchmarks for Kimi K2.5 and MiniMax M3
BenchmarkKimi K2.5MiniMax M3Leader
MMLU Pro85.9%84.2%Kimi K2.5
LegalBench (Vals AI)Not measured85.4%Only MiniMax M3 measured
CaseLaw (Vals AI)58.7%Not measuredOnly Kimi K2.5 measured
CorpFin (Vals AI)68.3%68.1%Kimi K2.5
TaxEval (Vals AI)74.2%72.7%Kimi K2.5
MortgageTax (Vals AI)66.5%68.4%MiniMax M3
MedQA (Vals AI)94.4%Not measuredOnly Kimi K2.5 measured
MedCode (Vals AI)39.3%46.3%MiniMax M3
SimpleQA Verified34.3%Not measuredOnly Kimi K2.5 measured
LMArena Text: science14591452Kimi K2.5
LMArena Text: medicine and healthcare14481441Kimi K2.5
LMArena Text: legal and government14411437Kimi K2.5
LMArena Text: business and finance14271434MiniMax M3
PRBench Finance46.5%Not measuredOnly Kimi K2.5 measured
PRBench Legal43.8%Not measuredOnly Kimi K2.5 measured
Neither model has a result on the other 5 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index17
Web design benchmarks for Kimi K2.5 and MiniMax M3
BenchmarkKimi K2.5MiniMax M3Leader
Design Arena: websites12481257MiniMax M3
Design Arena: UI components12341243MiniMax M3
Design Arena: game dev12061210MiniMax M3
Design Arena: data viz12191236MiniMax M3
Design Arena: 3D12041208MiniMax M3
Design Arena: all code categories12361246MiniMax M3
Design Arena: SVG11481160MiniMax M3
Design Arena: ASCII art11651161Kimi K2.5
LiveBench designNot measured66.2%Only MiniMax M3 measured
GeneralLike-for-likePublic boards, shown as published, kept out of the index226
General benchmarks for Kimi K2.5 and MiniMax M3
BenchmarkKimi K2.5MiniMax M3Leader
LiveBench data analysisNot measured76.2%Only MiniMax M3 measured
Public Benefits Bench (Vals AI)Not measured64.1%Only MiniMax M3 measured
SAGE (Vals AI)49.9%50.6%MiniMax M3
MedScribe (Vals AI)76.4%87.3%MiniMax M3
Vals Multimodal IndexNot measured60.0%Only MiniMax M3 measured
IFBench70.2%82.9%MiniMax M3
Long context reasoning (AA-LCR)78.0%83.0%MiniMax M3
DTBenchNot measured78.9%Only MiniMax M3 measured
LMCANot measured33.7%Only MiniMax M3 measured
Fiction.LiveBench86.1%Not measuredOnly Kimi K2.5 measured
CL-bench19.3%Not measuredOnly Kimi K2.5 measured
Surface Evolver BenchNot measured55.0%Only MiniMax M3 measured
CL-bench Life13.2%Not measuredOnly Kimi K2.5 measured
LMArena Text14451433Kimi K2.5
LMArena Text: instruction following14311433MiniMax M3
LMArena Text: creative writing14221404Kimi K2.5
LMArena Vision12691251Kimi K2.5
LMArena Text: English14531443Kimi K2.5
LMArena Text: non-English14331420Kimi K2.5
LMArena Text: Chinese14941466Kimi K2.5
LMArena Text: German14411428Kimi K2.5
LMArena Text: Japanese14231380Kimi K2.5
LMArena Text: Korean14121373Kimi K2.5
LMArena Text: French14541444Kimi K2.5
LMArena Text: Spanish14501433Kimi K2.5
LMArena Text: Russian14351426Kimi K2.5
LMArena Text: Polish14451423Kimi K2.5
LMArena Text: multi-turn14441443Kimi K2.5
LMArena Text: longer queries14451446MiniMax M3
LMArena Text: writing and language14251419Kimi K2.5
LMArena Text: entertainment and media14171403Kimi K2.5
LMArena Vision: English12741263Kimi K2.5
LMArena Vision: Chinese13281311Kimi K2.5
LMArena Vision: reading text12761259Kimi K2.5
LMArena Vision: humor12841235Kimi K2.5
LMArena Vision: captioning1257Not measuredOnly Kimi K2.5 measured
LMArena Vision: recognizing things12851197Kimi K2.5
LMArena Vision: creative writing12691234Kimi K2.5
LMArena DocumentNot measured1435Only MiniMax M3 measured
MultiChallenge61.4%Not measuredOnly Kimi K2.5 measured
EQ-Bench 4Not measured1150Only MiniMax M3 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 MiniMax M3
BenchmarkKimi K2.5MiniMax M3Leader
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$0.0020Fits in one request
MiniMax M3$0.0009Fits in one request

MiniMax M3 costs less

Repository review

50K input + 3K output tokens

Kimi K2.5$0.035Fits in one request
MiniMax M3$0.019Fits in one request

MiniMax M3 costs less

Agent loop

200K input + 10K output tokens

Kimi K2.5$0.135Fits in one request
MiniMax M3$0.072Fits in one request

MiniMax M3 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 MiniMax M3
SpecificationMoonshotAI Kimi K2.5MiniMax M3
MakerMoonshot AIMiniMax
WeightsOpen weightsOpen weights
ReleasedJanuary 27, 2026May 31, 2026
Context window262K1.05M
Input $/Mtok$0.532$0.30
Output $/Mtok$2.85$1.20
Coding index38.639.2
Independent measurements behind it36
Cost per solved task$0.24$0.56
SourceOpenRouter listingOpenRouter listing

Questions

Is Kimi K2.5 better than MiniMax M3 for coding?

Kimi K2.5 and MiniMax M3 are level for coding, 38.6 against 39.2 on the Vibe Coding Index, so price and context decide. Pick Kimi K2.5 for a lower cost per solved task ($0.24 against $0.56) and work measured by LiveCodeBench (Vals AI) or ALE-Bench, where it leads. Pick MiniMax M3 for output tokens 2.4x cheaper ($1.20 vs $2.85 per million) and a larger context window (1.05M against 262K tokens).

Which is cheaper, Kimi K2.5 or MiniMax M3?

Kimi K2.5 costs $0.532 per million input tokens and $2.85 per million output tokens; MiniMax M3 costs $0.30 and $1.20. Per solved SWE-bench Verified task, Kimi K2.5 costs $0.24 and MiniMax M3 $0.56.

Which has the bigger context window, Kimi K2.5 or MiniMax M3?

MiniMax M3, with 1.05M tokens against 262K.

How many benchmarks is the Kimi K2.5 vs MiniMax M3 comparison based on?

82 benchmarks have a result for both models as of October 22, 2026. Kimi K2.5's coding index rests on 3 independent benchmarks and MiniMax M3's on 6. Where only one model has a result, the page says so instead of naming a winner.

Compare next

Embed this comparison on your site

Free under CC BY 4.0, while the source link stays visible. Preview the frame · Terms and other embeds

<iframe src="https://llmbenchmark.ai/embed/compare/kimi-k2-5-vs-minimax-minimax-m3" title="Kimi K2.5 vs MiniMax M3, LLMBenchmark.ai" width="100%" height="360" style="border:0;max-width:100%" loading="lazy"></iframe>
<p style="font:13px/1.4 sans-serif;margin:6px 0 0">Source: <a href="https://llmbenchmark.ai/compare/kimi-k2-5-vs-minimax-minimax-m3">LLMBenchmark.ai, Kimi K2.5 vs MiniMax M3</a></p>