Moonshot AI

Kimi K3

Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work

moonshotai/kimi-k3

Context

1M

Output

131.1K

Released

2026-07-16

Knowledge cutoff

—

Types

InputTextImageVideo
OutputText

Reasoning

Supported

Tool use

Supported

Structured output

Supported

Attachments

Supported

Open weights

Supported

Provider pricing

Nvidia
InTextImageVideo
OutText
Price
Free / Free
Context
1M
Status
Stable
OpenRouter
InTextImageVideo
OutText
Price
$0.30 / $11.94
Context
1M
Status
Stable
NanoGPT
InTextImageVideo
OutText
Price
$2.00 / $10.00
Context
1M
Status
Stable
IteraCompute
InTextImage
OutText
Price
$2.39 / $11.99
Context
1M
Status
Stable
Wallaby
InTextImageVideo
OutText
Price
$2.70 / $13.50
Context
1M
Status
Stable
ai&
InTextImagePDF
OutText
Price
$3.00 / $12.50
Context
1M
Status
Stable
Arcee
InText
OutText
Price
$3.00 / $15.00
Context
1M
Status
beta
Cloudflare AI Gateway
InTextImageVideo
OutText
Price
$3.00 / $15.00
Context
1M
Status
Stable
Impossibl
InTextImageVideo
OutText
Price
$3.00 / $15.00
Context
1M
Status
Stable
Kilo Gateway
InTextImageVideo
OutText
Price
$3.00 / $15.00
Context
1M
Status
Stable
Novita AI
InTextImageVideo
OutText
Price
$3.00 / $15.00
Context
1M
Status
Stable
Ofox
InTextImageVideo
OutText
Price
$3.00 / $15.00
Context
1M
Status
Stable
TensorX
InTextImageVideo
OutText
Price
$3.00 / $15.00
Context
1M
Status
Stable
TokenGo
InTextImageVideo
OutText
Price
$3.00 / $15.00
Context
1M
Status
Stable
Vercel AI Gateway
InTextImagePDF
OutText
Price
$3.00 / $15.00
Context
1M
Status
Stable
ZenMux
InTextImageVideo
OutText
Price
$3.00 / $15.00
Context
1M
Status
Stable
NEAR AI Cloud
InText
OutText
Price
$3.30 / $16.50
Context
1M
Status
Stable
Prime Inference
InTextImageVideo
OutText
Price
$3.45 / $17.25
Context
1M
Status
Stable

Benchmarks

  • DeepSWE

    67.5 · resolve rate

  • Terminal-Bench

    88.3 · accuracy

  • FrontierSWE

    81.2 · dominance score

  • Program Bench

    77.8 · score

  • SWE Marathon

    42 · resolve rate

  • GDPval-AA

    1668 · Elo

  • AA-Briefcase

    1548 · Elo

  • AutomationBench

    30.8 · success rate

  • JobBench

    52.9 · score

  • SpreadsheetBench

    34.8 · score

  • BrowseComp

    91.2 · accuracy

  • CharXiv Reasoning

    91.3 · accuracy

  • ZeroBench

    41 · pass@5

From the same lab