Enter opens the highlighted result. Esc closes.

GPT-5.4 vs Kimi K2.5

GPT-5.4 leads Kimi K2.5 on coding, 64.4 to 38.6 on the Vibe Coding Index, and Kimi K2.5 costs 5.3x less per output token ($2.85 vs $15.00 per million). Updated October 22, 2026; 89 benchmarks have a result for both.

Replace GPT-5.4 Replace Kimi K2.5
GPT-5.4OpenAI
Coding
64.4
Overall
69.6
Output $/M
$15.00
70
Kimi K2.5Moonshot AI
Coding
38.6
Overall
44.8
Output $/M
$2.85
19
vs boards won of 89 shared

GPT-5.4 or Kimi K2.5: which to pick

Pick GPT-5.4 for coding: it scores 64.4 on the Vibe Coding Index against 38.6 for Kimi K2.5.

Pick Kimi K2.5 for output tokens 5.3x cheaper ($2.85 vs $15.00 per million) and a lower cost per solved task ($0.24 against $1.02).

Model AGPT-5.4OpenAI64.4/1003 independent benchmarks$1.02 per solved taskModel BKimi K2.5Moonshot AI38.6/1003 independent benchmarks$0.24 per solved task
Shared results
89
GPT-5.4 only
21
Kimi K2.5 only
10
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
GPT-5.4 64.4 from 3 benchmarks, Kimi K2.5 38.6 from 3 benchmarks.
Widest category
Coding, 16.9 points apart on average over 15 shared benchmarks.
Widest single gap
ARC-AGI-2: GPT-5.4 74.0% against Kimi K2.5 11.8%, 62.1 points apart on a 0 to 100 scale.
Output price
GPT-5.4 $15.00 against Kimi K2.5 $2.85 per million output tokens, 5.3x apart.
Context window
GPT-5.4 takes 1.05M tokens, Kimi K2.5 262K.

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
CategoryGPT-5.4BarsKimi K2.5Ties
Coding141
Intelligence20
Agentic work16
Reasoning111
Math30
Knowledge102
Web design17
General282
All shared boards7019
CodingLike-for-likeSix of these boards feed the Vibe Coding Index141
Coding benchmarks for GPT-5.4 and Kimi K2.5
BenchmarkGPT-5.4Kimi K2.5Leader
SWE-bench VerifiedNot measured70.8%Only Kimi K2.5 measured
SWE-bench MultilingualNot measured67.3%Only Kimi K2.5 measured
LMArena WebDev14641437GPT-5.4
AA Coding Index71.146.8GPT-5.4
LiveCodeBench (Vals AI)84.1%83.9%GPT-5.4
SWE-bench Verified (Vals AI)78.2%70.0%GPT-5.4
Vibe Code Bench (Vals AI)67.4%17.5%GPT-5.4
Code migration (Vals AI)35.0%7.0%GPT-5.4
ProgramBench (Vals AI)0.5%Not measuredOnly GPT-5.4 measured
Terminal-Bench 2.1 (Vals AI)Not measured42.0%Only Kimi K2.5 measured
SkillsBench (Vals AI)51.7%Not measuredOnly GPT-5.4 measured
Terminal-Bench Hard (Artificial Analysis run)57.6%34.8%GPT-5.4
Terminal-Bench81.8%43.2%GPT-5.4
GSO-Bench31.4%Not measuredOnly GPT-5.4 measured
WeirdML77.7%45.6%GPT-5.4
MirrorCode15.6%Not measuredOnly GPT-5.4 measured
ALE-Bench1607.0821.6GPT-5.4
LMArena Text: coding14961483GPT-5.4
LMArena Text: software and IT14911471GPT-5.4
LMArena WebDev: React14541429GPT-5.4
LMArena WebDev: HTML14781426GPT-5.4
LMArena WebDev: from an image14361439Kimi K2.5
Neither model has a result on the other 17 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index20
Intelligence benchmarks for GPT-5.4 and Kimi K2.5
BenchmarkGPT-5.4Kimi K2.5Leader
AA Intelligence Index39.023.5GPT-5.4
Epoch Capabilities Index157.3148.1GPT-5.4
Agentic workLike-for-likePublic boards, shown as published, kept out of the index16
Agentic work benchmarks for GPT-5.4 and Kimi K2.5
BenchmarkGPT-5.4Kimi K2.5Leader
Design Arena: agents building web apps10591136Kimi K2.5
Design Arena: agents building full-stack apps10171114Kimi K2.5
Design Arena: agents building mobile apps10871147Kimi K2.5
Design Arena: agents building Android apps10311118Kimi K2.5
Design Arena: agents building Godot games11351219Kimi K2.5
Finance Agent (Vals AI)Not measured35.8%Only Kimi K2.5 measured
Legal Research Bench (Vals AI)Not measured15.9%Only Kimi K2.5 measured
Excel Modeling Benchmark (Vals AI)Not measured28.5%Only Kimi K2.5 measured
τ²-Bench Telecom87.1%95.9%Kimi K2.5
APEX-Agents52.4%Not measuredOnly GPT-5.4 measured
METR time horizons74.3%Not measuredOnly GPT-5.4 measured
DeepResearch Bench35.1%Not measuredOnly GPT-5.4 measured
PostTrainBench19.0%Not measuredOnly GPT-5.4 measured
OSWorldNot measured63.3%Only Kimi K2.5 measured
Vending-Bench 2$6,144$1,198GPT-5.4
Neither model has a result on the other 26 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index111
Reasoning benchmarks for GPT-5.4 and Kimi K2.5
BenchmarkGPT-5.4Kimi K2.5Leader
GPQA Diamond91.7%84.1%GPT-5.4
MMMU Pro (Vals AI)87.5%84.3%GPT-5.4
GPQA Diamond (Artificial Analysis run)92.0%87.9%GPT-5.4
Humanity's Last Exam (Artificial Analysis run)43.7%30.7%GPT-5.4
GPQA Diamond (Epoch AI run)93.3%Not measuredOnly GPT-5.4 measured
ARC-AGI93.7%65.3%GPT-5.4
ARC-AGI-274.0%11.8%GPT-5.4
SimpleBenchNot measured46.8%Only Kimi K2.5 measured
Chess puzzles48.0%Not measuredOnly GPT-5.4 measured
Mystery game puzzles37.0%Not measuredOnly GPT-5.4 measured
EBR-bench25.4%Not measuredOnly GPT-5.4 measured
LMArena Text: hard prompts14851453GPT-5.4
LMArena Text: expert15071466GPT-5.4
LMArena Text: hard prompts in English14791460GPT-5.4
LMArena Vision: diagrams13271282GPT-5.4
LMArena Vision: homework13361297GPT-5.4
Kagi LLM Benchmark63.8%78.5%Kimi K2.5
Neither model has a result on the other 5 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index30
Math benchmarks for GPT-5.4 and Kimi K2.5
BenchmarkGPT-5.4Kimi K2.5Leader
AIME96.7%95.6%GPT-5.4
OTIS Mock AIME97.8%Not measuredOnly GPT-5.4 measured
FrontierMath, tiers 1 to 378.6%Not measuredOnly GPT-5.4 measured
FrontierMath, tier 449.0%Not measuredOnly GPT-5.4 measured
ProofBench56.0%Not measuredOnly GPT-5.4 measured
LMArena Text: math14881470GPT-5.4
LMArena Text: mathematical work14971473GPT-5.4
Neither model has a result on the other 6 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index102
Knowledge benchmarks for GPT-5.4 and Kimi K2.5
BenchmarkGPT-5.4Kimi K2.5Leader
MMLU Pro87.5%85.9%GPT-5.4
LegalBench (Vals AI)86.0%Not measuredOnly GPT-5.4 measured
CaseLaw (Vals AI)63.8%58.7%GPT-5.4
CorpFin (Vals AI)65.3%68.3%Kimi K2.5
TaxEval (Vals AI)74.0%74.2%Kimi K2.5
MortgageTax (Vals AI)68.3%66.5%GPT-5.4
MedQA (Vals AI)96.1%94.4%GPT-5.4
MedCode (Vals AI)41.3%39.3%GPT-5.4
SimpleQA Verified45.1%34.3%GPT-5.4
LMArena Text: science14791459GPT-5.4
LMArena Text: medicine and healthcare14581448GPT-5.4
LMArena Text: legal and government14861441GPT-5.4
LMArena Text: business and finance14741427GPT-5.4
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 GPT-5.4 and Kimi K2.5
BenchmarkGPT-5.4Kimi K2.5Leader
Design Arena: websites11981248Kimi K2.5
Design Arena: UI components12121234Kimi K2.5
Design Arena: game dev11941206Kimi K2.5
Design Arena: data viz11961219Kimi K2.5
Design Arena: 3D10651204Kimi K2.5
Design Arena: all code categories11791236Kimi K2.5
Design Arena: SVG11411149Kimi K2.5
Design Arena: ASCII art11881166GPT-5.4
Neither model has a result on the other 1 board here
GeneralLike-for-likePublic boards, shown as published, kept out of the index282
General benchmarks for GPT-5.4 and Kimi K2.5
BenchmarkGPT-5.4Kimi K2.5Leader
SAGE (Vals AI)43.3%49.9%Kimi K2.5
MedScribe (Vals AI)77.5%76.4%GPT-5.4
IFBench73.9%70.2%GPT-5.4
Long context reasoning (AA-LCR)82.0%78.0%GPT-5.4
DTBench94.4%Not measuredOnly GPT-5.4 measured
LMCA52.0%Not measuredOnly GPT-5.4 measured
Fiction.LiveBenchNot measured86.1%Only Kimi K2.5 measured
CL-bench27.9%19.3%GPT-5.4
Furniture assembly37.5%Not measuredOnly GPT-5.4 measured
CL-bench Life21.7%13.2%GPT-5.4
LMArena Text14691445GPT-5.4
LMArena Text: instruction following14691431GPT-5.4
LMArena Text: creative writing14391422GPT-5.4
LMArena Vision13031269GPT-5.4
LMArena Text: English14671453GPT-5.4
LMArena Text: non-English14651433GPT-5.4
LMArena Text: Chinese15201494GPT-5.4
LMArena Text: German14721441GPT-5.4
LMArena Text: Japanese14851423GPT-5.4
LMArena Text: Korean14491412GPT-5.4
LMArena Text: French14931454GPT-5.4
LMArena Text: Spanish14521450GPT-5.4
LMArena Text: Russian14801435GPT-5.4
LMArena Text: Polish14721445GPT-5.4
LMArena Text: multi-turn14821444GPT-5.4
LMArena Text: longer queries14731445GPT-5.4
LMArena Text: writing and language14581425GPT-5.4
LMArena Text: entertainment and media14381417GPT-5.4
LMArena Vision: English13011274GPT-5.4
LMArena Vision: Chinese13541328GPT-5.4
LMArena Vision: reading text13121276GPT-5.4
LMArena Vision: humor12841284GPT-5.4
LMArena Vision: captioningNot measured1257Only Kimi K2.5 measured
LMArena Vision: recognizing things12511285Kimi K2.5
LMArena Vision: creative writing12831269GPT-5.4
LMArena Document1471Not measuredOnly GPT-5.4 measured
Neither model has a result on the other 10 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for GPT-5.4 and Kimi K2.5
BenchmarkGPT-5.4Kimi K2.5Leader
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

GPT-5.4$0.010Fits in one request
Kimi K2.5$0.0020Fits in one request

Kimi K2.5 costs less

Repository review

50K input + 3K output tokens

GPT-5.4$0.170Fits in one request
Kimi K2.5$0.035Fits in one request

Kimi K2.5 costs less

Agent loop

200K input + 10K output tokens

GPT-5.4$0.650Fits in one request
Kimi K2.5$0.135Fits 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 GPT-5.4 and Kimi K2.5
SpecificationOpenAI GPT-5.4MoonshotAI Kimi K2.5
MakerOpenAIMoonshot AI
WeightsNo public weights listedOpen weights
ReleasedMarch 5, 2026January 27, 2026
Context window1.05M262K
Input $/Mtok$2.50$0.532
Output $/Mtok$15.00$2.85
Coding index64.438.6
Independent measurements behind it33
Cost per solved task$1.02$0.24
SourceOpenRouter listingOpenRouter listing

Questions

Is GPT-5.4 better than Kimi K2.5 for coding?

Pick GPT-5.4 for coding: it scores 64.4 on the Vibe Coding Index against 38.6 for Kimi K2.5. Pick Kimi K2.5 for output tokens 5.3x cheaper ($2.85 vs $15.00 per million) and a lower cost per solved task ($0.24 against $1.02).

Which is cheaper, GPT-5.4 or Kimi K2.5?

GPT-5.4 costs $2.50 per million input tokens and $15.00 per million output tokens; Kimi K2.5 costs $0.532 and $2.85. Per solved SWE-bench Verified task, GPT-5.4 costs $1.02 and Kimi K2.5 $0.24.

Which has the bigger context window, GPT-5.4 or Kimi K2.5?

GPT-5.4, with 1.05M tokens against 262K.

How many benchmarks is the GPT-5.4 vs Kimi K2.5 comparison based on?

89 benchmarks have a result for both models as of October 22, 2026. GPT-5.4's coding index rests on 3 independent benchmarks and Kimi K2.5's on 3. 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/gpt-5-4-vs-kimi-k2-5" title="GPT-5.4 vs Kimi K2.5, 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/gpt-5-4-vs-kimi-k2-5">LLMBenchmark.ai, GPT-5.4 vs Kimi K2.5</a></p>