Enter opens the highlighted result. Esc closes.

Gemini 3 Flash Preview vs Kimi K2 Thinking

Updated October 21, 2026. 27 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.

Replace Gemini 3 Flash Preview Replace Kimi K2 Thinking
Gemini 3 Flash PreviewGoogle
Coding
75.8
Overall
26.3
Output $/M
$3.00
25
Kimi K2 ThinkingMoonshot AI
Coding
63.4
Overall
22.0
Output $/M
$2.50
2
vs boards won of 27 shared

Decision reading

Pick Gemini 3 Flash Preview for coding: it scores 75.8 on the Vibe Coding Index against 63.4 for Kimi K2 Thinking.

Pick Kimi K2 Thinking when output tokens cost 1.2x less, or when what matters is τ²-Bench Telecom or CaseLaw (Vals AI), where it leads.

Model AGemini 3 Flash PreviewGoogle75.8/1001 independent benchmark$0.37 per solved taskModel BKimi K2 ThinkingMoonshot AI63.4/1001 independent benchmark$0.61 per solved task
Shared results
27
Gemini 3 Flash Preview only
42
Kimi K2 Thinking only
3
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 Flash Preview 75.8 from 1 benchmark, Kimi K2 Thinking 63.4 from 1 benchmark.
Widest category
Coding, 17.0 points apart on average over 8 shared benchmarks.
Widest single gap
Terminal-Bench: Gemini 3 Flash Preview 64.3% against Kimi K2 Thinking 35.7%, 28.5 points apart on a 0 to 100 scale.
Output price
Gemini 3 Flash Preview $3.00 against Kimi K2 Thinking $2.50 per million output tokens, 1.2x apart.
Context window
Gemini 3 Flash Preview takes 1.05M tokens, Kimi K2 Thinking 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
CategoryGemini 3 Flash PreviewBarsKimi K2 ThinkingTies
Coding80
Intelligence10
Agentic work01
Reasoning30
Math40
Knowledge61
Web design10
General20
All shared boards252
CodingLike-for-likeFour of these boards feed the Vibe Coding Index80
Coding benchmarks for Gemini 3 Flash Preview and Kimi K2 Thinking
BenchmarkGemini 3 Flash PreviewKimi K2 ThinkingLeader
SWE-bench Verified75.8%63.4%Gemini 3 Flash Preview
LiveCodeBench (Vals AI)85.6%63.1%Gemini 3 Flash Preview
SWE-bench Verified (Vals AI)75.0%60.2%Gemini 3 Flash Preview
Vibe Code Bench (Vals AI)20.2%Not measuredOnly Gemini 3 Flash Preview measured
Code migration (Vals AI)6.4%Not measuredOnly Gemini 3 Flash Preview measured
ProgramBench (Vals AI)0.0%Not measuredOnly Gemini 3 Flash Preview measured
Terminal-Bench 2.1 (Vals AI)53.9%Not measuredOnly Gemini 3 Flash Preview measured
LiveCodeBench (Artificial Analysis run)90.8%85.3%Gemini 3 Flash Preview
Terminal-Bench Hard (Artificial Analysis run)38.6%31.1%Gemini 3 Flash Preview
Terminal-Bench64.3%35.7%Gemini 3 Flash Preview
GSO-Bench9.8%Not measuredOnly Gemini 3 Flash Preview measured
WeirdML61.6%42.8%Gemini 3 Flash Preview
ALE-Bench1367.2597.5Gemini 3 Flash Preview
Neither model has a result on the other 25 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index10
Intelligence benchmarks for Gemini 3 Flash Preview and Kimi K2 Thinking
BenchmarkGemini 3 Flash PreviewKimi K2 ThinkingLeader
AA Intelligence Index26.322.0Gemini 3 Flash Preview
Epoch Capabilities IndexNot measured145.8Only Kimi K2 Thinking measured
Agentic workLike-for-likePublic boards, shown as published, kept out of the index01
Agentic work benchmarks for Gemini 3 Flash Preview and Kimi K2 Thinking
BenchmarkGemini 3 Flash PreviewKimi K2 ThinkingLeader
Design Arena: agents building web apps1111Not measuredOnly Gemini 3 Flash Preview measured
Design Arena: agents building full-stack apps1063Not measuredOnly Gemini 3 Flash Preview measured
Design Arena: agents building mobile apps1090Not measuredOnly Gemini 3 Flash Preview measured
Design Arena: agents building Android apps1056Not measuredOnly Gemini 3 Flash Preview measured
Design Arena: agents building Godot games1161Not measuredOnly Gemini 3 Flash Preview measured
Design Arena: PowerPoint slides1036Not measuredOnly Gemini 3 Flash Preview measured
Finance Agent (Vals AI)42.5%Not measuredOnly Gemini 3 Flash Preview measured
Tax Agent Bench (Vals AI)11.8%Not measuredOnly Gemini 3 Flash Preview measured
Legal Research Bench (Vals AI)18.3%Not measuredOnly Gemini 3 Flash Preview measured
Excel Modeling Benchmark (Vals AI)26.0%Not measuredOnly Gemini 3 Flash Preview measured
τ²-Bench Telecom80.4%93.0%Kimi K2 Thinking
METR time horizonsNot measured59.2%Only Kimi K2 Thinking measured
BALROG48.1%Not measuredOnly Gemini 3 Flash Preview measured
DeepResearch Bench49.8%Not measuredOnly Gemini 3 Flash Preview measured
Vending-Bench 2$3,635Not measuredOnly Gemini 3 Flash Preview measured
Neither model has a result on the other 24 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index30
Reasoning benchmarks for Gemini 3 Flash Preview and Kimi K2 Thinking
BenchmarkGemini 3 Flash PreviewKimi K2 ThinkingLeader
GPQA Diamond87.9%78.5%Gemini 3 Flash Preview
MMMU Pro (Vals AI)87.6%Not measuredOnly Gemini 3 Flash Preview measured
GPQA Diamond (Artificial Analysis run)89.8%83.8%Gemini 3 Flash Preview
Humanity's Last Exam (Artificial Analysis run)36.6%23.8%Gemini 3 Flash Preview
GPQA Diamond (Epoch AI run)89.4%Not measuredOnly Gemini 3 Flash Preview measured
ARC-AGI84.7%Not measuredOnly Gemini 3 Flash Preview measured
ARC-AGI-233.6%Not measuredOnly Gemini 3 Flash Preview measured
SimpleBench61.1%Not measuredOnly Gemini 3 Flash Preview measured
Chess puzzles39.0%Not measuredOnly Gemini 3 Flash Preview measured
Mystery game puzzles26.0%Not measuredOnly Gemini 3 Flash Preview measured
VPCT72.6%Not measuredOnly Gemini 3 Flash Preview measured
Neither model has a result on the other 10 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index40
Math benchmarks for Gemini 3 Flash Preview and Kimi K2 Thinking
BenchmarkGemini 3 Flash PreviewKimi K2 ThinkingLeader
AIME95.6%85.4%Gemini 3 Flash Preview
MGSM (Vals AI)93.3%90.2%Gemini 3 Flash Preview
AA Math Index97.094.7Gemini 3 Flash Preview
AIME 2025 (Artificial Analysis run)97.0%94.7%Gemini 3 Flash Preview
OTIS Mock AIME95.6%Not measuredOnly Gemini 3 Flash Preview measured
FrontierMath, tiers 1 to 351.2%Not measuredOnly Gemini 3 Flash Preview measured
FrontierMath, tier 417.1%Not measuredOnly Gemini 3 Flash Preview measured
ProofBench15.0%Not measuredOnly Gemini 3 Flash Preview measured
Neither model has a result on the other 5 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index61
Knowledge benchmarks for Gemini 3 Flash Preview and Kimi K2 Thinking
BenchmarkGemini 3 Flash PreviewKimi K2 ThinkingLeader
MMLU Pro88.6%81.1%Gemini 3 Flash Preview
LegalBench (Vals AI)86.9%80.2%Gemini 3 Flash Preview
CaseLaw (Vals AI)55.8%65.7%Kimi K2 Thinking
CorpFin (Vals AI)66.4%60.6%Gemini 3 Flash Preview
TaxEval (Vals AI)73.9%71.7%Gemini 3 Flash Preview
MortgageTax (Vals AI)68.7%Not measuredOnly Gemini 3 Flash Preview measured
MedQA (Vals AI)95.8%92.6%Gemini 3 Flash Preview
MedCode (Vals AI)55.9%Not measuredOnly Gemini 3 Flash Preview measured
MMLU Pro (Artificial Analysis run)89.0%84.8%Gemini 3 Flash Preview
SimpleQA Verified66.8%Not measuredOnly Gemini 3 Flash Preview measured
GeoBench88.0%Not measuredOnly Gemini 3 Flash Preview measured
Neither model has a result on the other 6 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index10
Web design benchmarks for Gemini 3 Flash Preview and Kimi K2 Thinking
BenchmarkGemini 3 Flash PreviewKimi K2 ThinkingLeader
Design Arena: websites11961113Gemini 3 Flash Preview
Design Arena: game dev1166Not measuredOnly Gemini 3 Flash Preview measured
Design Arena: 3D1184Not measuredOnly Gemini 3 Flash Preview measured
Design Arena: all code categories1189Not measuredOnly Gemini 3 Flash Preview measured
Neither model has a result on the other 5 boards here
GeneralLike-for-likePublic boards, shown as published, kept out of the index20
General benchmarks for Gemini 3 Flash Preview and Kimi K2 Thinking
BenchmarkGemini 3 Flash PreviewKimi K2 ThinkingLeader
SAGE (Vals AI)51.8%Not measuredOnly Gemini 3 Flash Preview measured
MedScribe (Vals AI)69.9%Not measuredOnly Gemini 3 Flash Preview measured
Vals Multimodal Index52.2%Not measuredOnly Gemini 3 Flash Preview measured
IFBench78.0%68.1%Gemini 3 Flash Preview
Long context reasoning (AA-LCR)78.0%72.0%Gemini 3 Flash Preview
DTBench89.1%Not measuredOnly Gemini 3 Flash Preview measured
LMCA43.1%Not measuredOnly Gemini 3 Flash Preview measured
CL-benchNot measured17.6%Only Kimi K2 Thinking measured
Neither model has a result on the other 38 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for Gemini 3 Flash Preview and Kimi K2 Thinking
BenchmarkGemini 3 Flash PreviewKimi K2 ThinkingLeader
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 Flash Preview$0.0020Fits in one request
Kimi K2 Thinking$0.0019Fits in one request

Kimi K2 Thinking costs less

Repository review

50K input + 3K output tokens

Gemini 3 Flash Preview$0.034Fits in one request
Kimi K2 Thinking$0.037Fits in one request

Gemini 3 Flash Preview costs less

Agent loop

200K input + 10K output tokens

Gemini 3 Flash Preview$0.130Fits in one request
Kimi K2 Thinking$0.145Fits in one request

Gemini 3 Flash Preview 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 Flash Preview and Kimi K2 Thinking
SpecificationGoogle Gemini 3 Flash PreviewMoonshotAI Kimi K2 Thinking
MakerGoogleMoonshot AI
WeightsNo public weights listedOpen weights
ReleasedDecember 17, 2025November 6, 2025
Context window1.05M262K
Input $/Mtok$0.50$0.60
Output $/Mtok$3.00$2.50
Coding index75.863.4
Independent measurements behind it11
Cost per solved task$0.37$0.61
SourceOpenRouter listingOpenRouter listing

Questions

Is Gemini 3 Flash Preview better than Kimi K2 Thinking for coding?

Pick Gemini 3 Flash Preview for coding: it scores 75.8 on the Vibe Coding Index against 63.4 for Kimi K2 Thinking. Pick Kimi K2 Thinking when output tokens cost 1.2x less, or when what matters is τ²-Bench Telecom or CaseLaw (Vals AI), where it leads.

Which is cheaper, Gemini 3 Flash Preview or Kimi K2 Thinking?

Gemini 3 Flash Preview costs $0.50 per million input tokens and $3.00 per million output tokens; Kimi K2 Thinking costs $0.60 and $2.50. Per solved SWE-bench Verified task, Gemini 3 Flash Preview costs $0.37 and Kimi K2 Thinking $0.61.

Which has the bigger context window, Gemini 3 Flash Preview or Kimi K2 Thinking?

Gemini 3 Flash Preview, with 1.05M tokens against 262K.

How many benchmarks is the Gemini 3 Flash Preview vs Kimi K2 Thinking comparison based on?

27 benchmarks have a result for both models as of October 21, 2026. Gemini 3 Flash Preview's coding index rests on 1 independent benchmark and Kimi K2 Thinking's on 1. Where only one model has a result, the page says so instead of naming a winner.

Compare next