LLMBenchmark.ai

Enter opens the highlighted result. Esc closes.

Kimi K2.5 vs Qwen3.5 397B A17B

Updated October 21, 2026. 57 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 Qwen3.5 397B A17B
Kimi K2.5Moonshot AI
Coding
58.5
Overall
23.5
Output $/M
$2.50
46
Qwen3.5 397B A17BQwen
Coding
28.6
Overall
21.4
Output $/M
$2.34
11
vs boards won of 57 shared

Decision reading

Pick Kimi K2.5 for coding: it scores 58.5 on the Vibe Coding Index against 28.6 for Qwen3.5 397B A17B.

Pick Qwen3.5 397B A17B when output tokens cost 1.1x less, or when what matters is Terminal-Bench Hard (Artificial Analysis run) or GPQA Diamond (Artificial Analysis run), where it leads.

Model AKimi K2.5Moonshot AI58.5/1002 independent benchmarks$0.24 per solved taskModel BQwen3.5 397B A17BQwen28.6/1001 independent benchmarkNo cost per solved task
Shared results
57
Kimi K2.5 only
41
Qwen3.5 397B A17B only
8
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, Qwen3.5 397B A17B 28.6 from 1 benchmark.
Widest category
Web design, 6.3 points apart on average over 7 shared benchmarks.
Widest single gap
Design Arena: UI components: Kimi K2.5 1236 against Qwen3.5 397B A17B 1163, 9.1 points apart on a 0 to 100 scale.
Output price
Kimi K2.5 $2.50 against Qwen3.5 397B A17B $2.34 per million output tokens, 1.1x 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 K2.5BarsQwen3.5 397B A17BTies
Coding61
Intelligence20
Agentic work10
Reasoning61
Math20
Knowledge22
Web design70
General207
All shared boards4611
CodingLike-for-likeFour of these boards feed the Vibe Coding Index61
Coding benchmarks for Kimi K2.5 and Qwen3.5 397B A17B
BenchmarkKimi K2.5Qwen3.5 397B A17BLeader
SWE-bench Verified70.8%Not measuredOnly Kimi K2.5 measured
SWE-bench Multilingual67.3%Not measuredOnly Kimi K2.5 measured
LMArena WebDev14371400Kimi K2.5
AA Coding Index46.8Not measuredOnly Kimi K2.5 measured
LiveCodeBench (Vals AI)83.9%Not measuredOnly Kimi K2.5 measured
SWE-bench Verified (Vals AI)70.0%Not measuredOnly Kimi K2.5 measured
Vibe Code Bench (Vals AI)17.5%Not measuredOnly Kimi K2.5 measured
Code migration (Vals AI)7.0%Not measuredOnly Kimi K2.5 measured
Terminal-Bench 2.1 (Vals AI)42.0%Not measuredOnly Kimi K2.5 measured
SciCodeNot measured44.8%Only Qwen3.5 397B A17B measured
Terminal-Bench Hard (Artificial Analysis run)34.8%40.9%Qwen3.5 397B A17B
Terminal-Bench43.2%Not measuredOnly Kimi K2.5 measured
WeirdML45.6%Not measuredOnly Kimi K2.5 measured
ALE-Bench821.6621.9Kimi K2.5
LMArena Text: coding14831464Kimi K2.5
LMArena Text: software and IT14711463Kimi K2.5
LMArena WebDev: React14291388Kimi K2.5
LMArena WebDev: HTML14261410Kimi K2.5
LMArena WebDev: from an image1439Not measuredOnly Kimi K2.5 measured
Neither model has a result on the other 19 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index20
Intelligence benchmarks for Kimi K2.5 and Qwen3.5 397B A17B
BenchmarkKimi K2.5Qwen3.5 397B A17BLeader
AA Intelligence Index23.521.4Kimi K2.5
Epoch Capabilities Index148.1146.8Kimi K2.5
Agentic workLike-for-likePublic boards, shown as published, kept out of the index10
Agentic work benchmarks for Kimi K2.5 and Qwen3.5 397B A17B
BenchmarkKimi K2.5Qwen3.5 397B A17BLeader
Design Arena: agents building web apps1136Not measuredOnly Kimi K2.5 measured
Design Arena: agents building full-stack apps1114Not measuredOnly Kimi K2.5 measured
Design Arena: agents building mobile apps1147Not measuredOnly Kimi K2.5 measured
Design Arena: agents building Android apps1124Not measuredOnly Kimi K2.5 measured
Design Arena: agents building Godot games1219Not measuredOnly Kimi K2.5 measured
Finance Agent (Vals AI)35.8%Not measuredOnly Kimi K2.5 measured
Legal Research Bench (Vals AI)15.9%Not measuredOnly Kimi K2.5 measured
Excel Modeling Benchmark (Vals AI)28.5%Not measuredOnly Kimi K2.5 measured
τ²-Bench Telecom95.9%95.6%Kimi K2.5
APEX-AgentsNot measured24.9%Only Qwen3.5 397B A17B measured
OSWorld63.3%Not measuredOnly Kimi K2.5 measured
Vending-Bench 2$1,198Not measuredOnly Kimi K2.5 measured
Neither model has a result on the other 27 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index61
Reasoning benchmarks for Kimi K2.5 and Qwen3.5 397B A17B
BenchmarkKimi K2.5Qwen3.5 397B A17BLeader
GPQA Diamond84.1%Not measuredOnly Kimi K2.5 measured
MMMU Pro (Vals AI)84.3%Not measuredOnly Kimi K2.5 measured
GPQA Diamond (Artificial Analysis run)87.9%89.3%Qwen3.5 397B A17B
Humanity's Last Exam (Artificial Analysis run)30.7%29.0%Kimi K2.5
GPQA Diamond (Epoch AI run)Not measured86.4%Only Qwen3.5 397B A17B 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
Mystery game puzzlesNot measured18.0%Only Qwen3.5 397B A17B measured
LMArena Text: hard prompts14531448Kimi K2.5
LMArena Text: expert14661462Kimi K2.5
LMArena Text: hard prompts in English14601450Kimi K2.5
LMArena Vision: diagrams12821277Kimi K2.5
LMArena Vision: homework12971292Kimi K2.5
Neither model has a result on the other 7 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index20
Math benchmarks for Kimi K2.5 and Qwen3.5 397B A17B
BenchmarkKimi K2.5Qwen3.5 397B A17BLeader
AIME95.6%Not measuredOnly Kimi K2.5 measured
OTIS Mock AIMENot measured88.9%Only Qwen3.5 397B A17B measured
FrontierMath, tiers 1 to 3Not measured31.2%Only Qwen3.5 397B A17B measured
LMArena Text: math14701453Kimi K2.5
LMArena Text: mathematical work14731454Kimi K2.5
Neither model has a result on the other 8 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index22
Knowledge benchmarks for Kimi K2.5 and Qwen3.5 397B A17B
BenchmarkKimi K2.5Qwen3.5 397B A17BLeader
MMLU Pro85.9%Not measuredOnly Kimi K2.5 measured
CaseLaw (Vals AI)58.7%Not measuredOnly Kimi K2.5 measured
CorpFin (Vals AI)68.3%Not measuredOnly Kimi K2.5 measured
TaxEval (Vals AI)74.2%Not measuredOnly Kimi K2.5 measured
MortgageTax (Vals AI)66.5%Not measuredOnly Kimi K2.5 measured
MedQA (Vals AI)94.4%Not measuredOnly Kimi K2.5 measured
MedCode (Vals AI)39.3%Not measuredOnly Kimi K2.5 measured
SimpleQA Verified34.3%Not measuredOnly Kimi K2.5 measured
LMArena Text: science14591457Kimi K2.5
LMArena Text: medicine and healthcare14481455Qwen3.5 397B A17B
LMArena Text: legal and government14411441Kimi K2.5
LMArena Text: business and finance14271435Qwen3.5 397B A17B
Neither model has a result on the other 5 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index70
Web design benchmarks for Kimi K2.5 and Qwen3.5 397B A17B
BenchmarkKimi K2.5Qwen3.5 397B A17BLeader
Design Arena: websites12491191Kimi K2.5
Design Arena: UI components12361163Kimi K2.5
Design Arena: game dev12071141Kimi K2.5
Design Arena: data viz12191175Kimi K2.5
Design Arena: 3D12051161Kimi K2.5
Design Arena: all code categories12361180Kimi K2.5
Design Arena: SVG11491139Kimi K2.5
Design Arena: ASCII art1165Not measuredOnly Kimi K2.5 measured
Neither model has a result on the other 1 board here
GeneralLike-for-likePublic boards, shown as published, kept out of the index207
General benchmarks for Kimi K2.5 and Qwen3.5 397B A17B
BenchmarkKimi K2.5Qwen3.5 397B A17BLeader
SAGE (Vals AI)49.9%Not measuredOnly Kimi K2.5 measured
MedScribe (Vals AI)76.4%Not measuredOnly Kimi K2.5 measured
IFBench70.2%78.8%Qwen3.5 397B A17B
Long context reasoning (AA-LCR)78.0%77.3%Kimi K2.5
DTBenchNot measured87.5%Only Qwen3.5 397B A17B measured
LMCANot measured37.9%Only Qwen3.5 397B A17B measured
Fiction.LiveBench86.1%Not measuredOnly Kimi K2.5 measured
CL-bench19.3%Not measuredOnly Kimi K2.5 measured
CL-bench Life13.2%Not measuredOnly Kimi K2.5 measured
LMArena Text14451437Kimi K2.5
LMArena Text: instruction following14311424Kimi K2.5
LMArena Text: creative writing14221402Kimi K2.5
LMArena Vision12691262Kimi K2.5
LMArena Text: English14531443Kimi K2.5
LMArena Text: non-English14331430Kimi K2.5
LMArena Text: Chinese14941498Qwen3.5 397B A17B
LMArena Text: German14411445Qwen3.5 397B A17B
LMArena Text: Japanese14231429Qwen3.5 397B A17B
LMArena Text: Korean14121384Kimi K2.5
LMArena Text: French14541462Qwen3.5 397B A17B
LMArena Text: Spanish14501441Kimi K2.5
LMArena Text: Russian14351430Kimi K2.5
LMArena Text: Polish14451447Qwen3.5 397B A17B
LMArena Text: multi-turn14441446Qwen3.5 397B A17B
LMArena Text: longer queries14451441Kimi K2.5
LMArena Text: writing and language14251415Kimi K2.5
LMArena Text: entertainment and media14171399Kimi K2.5
LMArena Vision: English12741268Kimi K2.5
LMArena Vision: Chinese13281314Kimi K2.5
LMArena Vision: reading text12761271Kimi K2.5
LMArena Vision: humor12841262Kimi K2.5
LMArena Vision: captioning12571230Kimi K2.5
LMArena Vision: recognizing things12851265Kimi K2.5
LMArena Vision: creative writing12691260Kimi K2.5
Neither model has a result on the other 12 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 Qwen3.5 397B A17B
BenchmarkKimi K2.5Qwen3.5 397B A17BLeader
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
Qwen3.5 397B A17B$0.0016Fits in one request

Qwen3.5 397B A17B costs less

Repository review

50K input + 3K output tokens

Kimi K2.5$0.033Fits in one request
Qwen3.5 397B A17B$0.027Fits in one request

Qwen3.5 397B A17B costs less

Agent loop

200K input + 10K output tokens

Kimi K2.5$0.125Fits in one request
Qwen3.5 397B A17B$0.101Fits in one request

Qwen3.5 397B A17B 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 Qwen3.5 397B A17B
SpecificationMoonshotAI Kimi K2.5Qwen3.5 397B A17B
MakerMoonshot AIQwen
WeightsOpen weightsOpen weights
ReleasedJanuary 27, 2026February 16, 2026
Context window262K262K
Input $/Mtok$0.50$0.39
Output $/Mtok$2.50$2.34
Coding index58.528.6
Independent measurements behind it21
Cost per solved task$0.24Not computable
SourceOpenRouter listingOpenRouter listing

Questions

Is Kimi K2.5 better than Qwen3.5 397B A17B for coding?

Pick Kimi K2.5 for coding: it scores 58.5 on the Vibe Coding Index against 28.6 for Qwen3.5 397B A17B. Pick Qwen3.5 397B A17B when output tokens cost 1.1x less, or when what matters is Terminal-Bench Hard (Artificial Analysis run) or GPQA Diamond (Artificial Analysis run), where it leads.

Which is cheaper, Kimi K2.5 or Qwen3.5 397B A17B?

Kimi K2.5 costs $0.50 per million input tokens and $2.50 per million output tokens; Qwen3.5 397B A17B costs $0.39 and $2.34. Cost per solved task is known for Kimi K2.5 only.

Which has the bigger context window, Kimi K2.5 or Qwen3.5 397B A17B?

Neither: both take 262K tokens.

How many benchmarks is the Kimi K2.5 vs Qwen3.5 397B A17B comparison based on?

57 benchmarks have a result for both models as of October 21, 2026. Kimi K2.5's coding index rests on 2 independent benchmarks and Qwen3.5 397B A17B's on 1. Where only one model has a result, the page says so instead of naming a winner.

Compare next