LLMBenchmark.ai

Enter opens the highlighted result. Esc closes.

Gemini 3.6 Flash vs Kimi K3

Updated October 21, 2026. 97 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.

Replace Gemini 3.6 Flash Replace Kimi K3
Gemini 3.6 FlashGoogle
Coding
48.3
Overall
34.0
Output $/M
$3.75
26
Kimi K3Moonshot AI
Coding
61.8
Overall
43.6
Output $/M
$11.936
70
vs boards won of 97 shared, 1 tied

Decision reading

Pick Kimi K3 for coding: it scores 61.8 on the Vibe Coding Index against 48.3 for Gemini 3.6 Flash.

Pick Gemini 3.6 Flash when output tokens cost 3.2x less, or when what matters is LiveCodeBench (Vals AI) or Code migration (Vals AI), where it leads.

Model AGemini 3.6 FlashGoogle48.3/1001 independent benchmark$1.49 per solved taskModel BKimi K3Moonshot AI61.8/1002 independent benchmarks$0.81 per solved task
Shared results
97
Gemini 3.6 Flash only
10
Kimi K3 only
20
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
Gemini 3.6 Flash 48.3 from 1 benchmark, Kimi K3 61.8 from 2 benchmarks.
Widest category
Math, 14.8 points apart on average over 6 shared benchmarks.
Widest single gap
ProofBench: Gemini 3.6 Flash 36.0% against Kimi K3 87.0%, 51.0 points apart on a 0 to 100 scale.
Output price
Gemini 3.6 Flash $3.75 against Kimi K3 $11.936 per million output tokens, 3.2x apart.
Context window
Both take 1.05M 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
CategoryGemini 3.6 FlashBarsKimi K3Ties
Coding217
Intelligence02
Agentic work214
Reasoning561 tie
Math15
Knowledge56
Web design06
General1114
All shared boards26701 tie
CodingLike-for-likeFour of these boards feed the Vibe Coding Index217
Coding benchmarks for Gemini 3.6 Flash and Kimi K3
BenchmarkGemini 3.6 FlashKimi K3Leader
LiveBench codingNot measured59.3%Only Kimi K3 measured
LMArena WebDev15381655Kimi K3
AA Coding Index69.276.2Kimi K3
LiveCodeBench (Vals AI)88.1%87.2%Gemini 3.6 Flash
SWE-bench Verified (Vals AI)79.6%93.4%Kimi K3
Terminal-Bench 4 (Vals AI)Not measured17.2%Only Kimi K3 measured
Vibe Code Bench (Vals AI)64.0%85.0%Kimi K3
Code migration (Vals AI)30.9%16.1%Gemini 3.6 Flash
ProgramBench (Vals AI)0.0%2.0%Kimi K3
IOI (Vals AI)35.1%48.9%Kimi K3
Terminal-Bench 2.1 (Vals AI)73.8%80.9%Kimi K3
SWE-bench Multimodal (Vals AI)Not measured44.2%Only Kimi K3 measured
Vibe Code Bench 1-100 (Vals AI)Not measured18.2%Only Kimi K3 measured
CyberBench (Vals AI)45.4%75.2%Kimi K3
SciCode53.4%59.5%Kimi K3
FrontierCode34.4%44.2%Kimi K3
FrontierSWENot measured25.9%Only Kimi K3 measured
WeirdML56.1%82.6%Kimi K3
ALE-Bench715.51524.5Kimi K3
LMArena Text: coding14911508Kimi K3
LMArena Text: software and IT14921499Kimi K3
LMArena WebDev: React15281670Kimi K3
LMArena WebDev: HTML15471644Kimi K3
LMArena WebDev: from an image15451579Kimi K3
Neither model has a result on the other 14 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index02
Intelligence benchmarks for Gemini 3.6 Flash and Kimi K3
BenchmarkGemini 3.6 FlashKimi K3Leader
AA Intelligence Index34.043.6Kimi K3
Epoch Capabilities Index154.5157.9Kimi K3
Agentic workLike-for-likePublic boards, shown as published, kept out of the index214
Agentic work benchmarks for Gemini 3.6 Flash and Kimi K3
BenchmarkGemini 3.6 FlashKimi K3Leader
AA Agentic Index29.050.0Kimi K3
Design Arena: agents building web apps11971296Kimi K3
Design Arena: agents building full-stack apps11811315Kimi K3
Design Arena: agents building mobile apps11991267Kimi K3
Design Arena: agents building Android apps12081256Kimi K3
Design Arena: agents building Godot gamesNot measured1223Only Kimi K3 measured
Design Arena: agents building games11811252Kimi K3
Design Arena: HTML slides11481256Kimi K3
Design Arena: PowerPoint slides11461273Kimi K3
Design Arena: spreadsheets10171058Kimi K3
LiveBench researchNot measured79.0%Only Kimi K3 measured
LiveBench browserNot measured16.0%Only Kimi K3 measured
LiveBench long agenticNot measured42.8%Only Kimi K3 measured
Finance Agent (Vals AI)56.3%53.1%Gemini 3.6 Flash
Tax Agent Bench (Vals AI)18.0%33.4%Kimi K3
Legal Research Bench (Vals AI)25.0%46.1%Kimi K3
Harvey's Legal Agent Benchmark (Vals AI)Not measured12.9%Only Kimi K3 measured
Excel Modeling Benchmark (Vals AI)65.4%66.7%Kimi K3
BioMysteryBench (Vals AI)58.5%72.6%Kimi K3
Terminal-Bench-Science (Vals AI)4.3%1.4%Gemini 3.6 Flash
Time Horizon Index (Vals AI)Not measured10.5%Only Kimi K3 measured
APEX-Agents46.9%47.8%Kimi K3
PostTrainBenchNot measured32.0%Only Kimi K3 measured
Vending-Bench 2Not measured$5,165Only Kimi K3 measured
Neither model has a result on the other 15 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index56
Reasoning benchmarks for Gemini 3.6 Flash and Kimi K3
BenchmarkGemini 3.6 FlashKimi K3Leader
GPQA Diamond93.4%92.9%Gemini 3.6 Flash
LiveBench reasoningNot measured90.7%Only Kimi K3 measured
MMMU Pro (Vals AI)88.4%88.2%Gemini 3.6 Flash
GPQA Diamond (Artificial Analysis run)92.8%93.5%Kimi K3
Humanity's Last Exam (Artificial Analysis run)40.8%46.9%Kimi K3
GPQA Diamond (Epoch AI run)94.1%93.1%Gemini 3.6 Flash
ARC-AGI91.2%94.5%Kimi K3
ARC-AGI-260.4%60.4%Tie
SimpleBenchNot measured60.7%Only Kimi K3 measured
Chess puzzles42.0%37.0%Gemini 3.6 Flash
Mystery game puzzles30.0%26.0%Gemini 3.6 Flash
LMArena Text: hard prompts14861495Kimi K3
LMArena Text: expert14901520Kimi K3
LMArena Text: hard prompts in English14841496Kimi K3
LMArena Vision: diagrams1314Not measuredOnly Gemini 3.6 Flash measured
LMArena Vision: homework1314Not measuredOnly Gemini 3.6 Flash measured
Neither model has a result on the other 5 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index15
Math benchmarks for Gemini 3.6 Flash and Kimi K3
BenchmarkGemini 3.6 FlashKimi K3Leader
OTIS Mock AIME94.2%97.2%Kimi K3
FrontierMath, tiers 1 to 358.9%72.2%Kimi K3
FrontierMath, tier 422.0%39.0%Kimi K3
ProofBench36.0%87.0%Kimi K3
LMArena Text: math15081494Gemini 3.6 Flash
LMArena Text: mathematical work14921505Kimi K3
Neither model has a result on the other 7 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index56
Knowledge benchmarks for Gemini 3.6 Flash and Kimi K3
BenchmarkGemini 3.6 FlashKimi K3Leader
MMLU Pro89.3%88.0%Gemini 3.6 Flash
LegalBench (Vals AI)86.7%86.2%Gemini 3.6 Flash
CorpFin (Vals AI)63.3%71.6%Kimi K3
TaxEval (Vals AI)74.9%75.7%Kimi K3
MortgageTax (Vals AI)67.6%66.3%Gemini 3.6 Flash
MedCode (Vals AI)53.1%49.4%Gemini 3.6 Flash
SimpleQA Verified66.2%50.6%Gemini 3.6 Flash
LMArena Text: science14891508Kimi K3
LMArena Text: medicine and healthcare14691489Kimi K3
LMArena Text: legal and government14791503Kimi K3
LMArena Text: business and finance14641470Kimi K3
Neither model has a result on the other 6 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index06
Web design benchmarks for Gemini 3.6 Flash and Kimi K3
BenchmarkGemini 3.6 FlashKimi K3Leader
Design Arena: websites13011336Kimi K3
Design Arena: UI components12971345Kimi K3
Design Arena: game dev12631361Kimi K3
Design Arena: data viz12891350Kimi K3
Design Arena: 3D12721388Kimi K3
Design Arena: all code categories12901361Kimi K3
Design Arena: SVGNot measured1301Only Kimi K3 measured
Design Arena: ASCII art1281Not measuredOnly Gemini 3.6 Flash measured
LiveBench designNot measured77.2%Only Kimi K3 measured
GeneralLike-for-likePublic boards, shown as published, kept out of the index1114
General benchmarks for Gemini 3.6 Flash and Kimi K3
BenchmarkGemini 3.6 FlashKimi K3Leader
Vals IndexNot measured53.0%Only Kimi K3 measured
LiveBench data analysisNot measured78.7%Only Kimi K3 measured
Public Benefits Bench (Vals AI)56.6%68.2%Kimi K3
SAGE (Vals AI)48.7%52.8%Kimi K3
MedScribe (Vals AI)79.7%88.0%Kimi K3
Vals Multimodal Index65.1%73.4%Kimi K3
Long context reasoning (AA-LCR)80.0%88.7%Kimi K3
DTBench95.5%91.2%Gemini 3.6 Flash
LMCA44.9%52.7%Kimi K3
Surface Evolver BenchNot measured95.0%Only Kimi K3 measured
Furniture assembly23.3%34.2%Kimi K3
LMArena Text14801476Gemini 3.6 Flash
LMArena Text: instruction following14671485Kimi K3
LMArena Text: creative writing14661457Gemini 3.6 Flash
LMArena Vision1299Not measuredOnly Gemini 3.6 Flash measured
LMArena Text: English14821481Gemini 3.6 Flash
LMArena Text: non-English14701466Gemini 3.6 Flash
LMArena Text: Chinese15311529Gemini 3.6 Flash
LMArena Text: German14801490Kimi K3
LMArena Text: Japanese14751494Kimi K3
LMArena Text: Korean14291460Kimi K3
LMArena Text: French15031492Gemini 3.6 Flash
LMArena Text: Spanish14761472Gemini 3.6 Flash
LMArena Text: Russian14881485Gemini 3.6 Flash
LMArena Text: Polish14731458Gemini 3.6 Flash
LMArena Text: multi-turn14811489Kimi K3
LMArena Text: longer queries14751494Kimi K3
LMArena Text: writing and language14701471Kimi K3
LMArena Text: entertainment and media14561453Gemini 3.6 Flash
LMArena Vision: English1295Not measuredOnly Gemini 3.6 Flash measured
LMArena Vision: Chinese1357Not measuredOnly Gemini 3.6 Flash measured
LMArena Vision: reading text1312Not measuredOnly Gemini 3.6 Flash measured
LMArena Vision: humor1289Not measuredOnly Gemini 3.6 Flash measured
LMArena Vision: creative writing1278Not measuredOnly Gemini 3.6 Flash measured
LMArena Document1456Not measuredOnly Gemini 3.6 Flash measured
Neither model has a result on the other 11 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for Gemini 3.6 Flash and Kimi K3
BenchmarkGemini 3.6 FlashKimi K3Leader
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

Gemini 3.6 Flash$0.0026Fits in one request
Kimi K3$0.0062Fits in one request

Gemini 3.6 Flash costs less

Repository review

50K input + 3K output tokens

Gemini 3.6 Flash$0.049Fits in one request
Kimi K3$0.048Fits in one request

Kimi K3 costs less

Agent loop

200K input + 10K output tokens

Gemini 3.6 Flash$0.188Fits in one request
Kimi K3$0.167Fits in one request

Kimi K3 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 Gemini 3.6 Flash and Kimi K3
SpecificationGoogle Gemini 3.6 FlashMoonshotAI Kimi K3
MakerGoogleMoonshot AI
WeightsNo public weights listedOpen weights
ReleasedJuly 21, 2026July 16, 2026
Context window1.05M1.05M
Input $/Mtok$0.75$0.239
Output $/Mtok$3.75$11.936
Coding index48.361.8
Independent measurements behind it12
Cost per solved task$1.49$0.81
SourceOpenRouter listingOpenRouter listing

Questions

Is Gemini 3.6 Flash better than Kimi K3 for coding?

Pick Kimi K3 for coding: it scores 61.8 on the Vibe Coding Index against 48.3 for Gemini 3.6 Flash. Pick Gemini 3.6 Flash when output tokens cost 3.2x less, or when what matters is LiveCodeBench (Vals AI) or Code migration (Vals AI), where it leads.

Which is cheaper, Gemini 3.6 Flash or Kimi K3?

Gemini 3.6 Flash costs $0.75 per million input tokens and $3.75 per million output tokens; Kimi K3 costs $0.239 and $11.936. Per solved SWE-bench Verified task, Gemini 3.6 Flash costs $1.49 and Kimi K3 $0.81.

Which has the bigger context window, Gemini 3.6 Flash or Kimi K3?

Neither: both take 1.05M tokens.

How many benchmarks is the Gemini 3.6 Flash vs Kimi K3 comparison based on?

97 benchmarks have a result for both models as of October 21, 2026. Gemini 3.6 Flash's coding index rests on 1 independent benchmark and Kimi K3's on 2. Where only one model has a result, the page says so instead of naming a winner.

Compare next