Kimi-K3

Kimi-K3

moonshotai/Kimi-K3

About Kimi-K3

Kimi K3 is Kimi’s most capable model to date, featuring 2.8 trillion parameters. Built on Kimi Delta Attention, a hybrid linear attention mechanism, and Attention Residuals, it natively supports visual understanding and offers a 1-million-token context window. It is designed for advanced AI applications such as software engineering, knowledge work, and deep reasoning.

Available Serverless

Run queries immediately, pay only for usage

Input Price

$

3.0

/ M Tokens

Cache Read

$

0.3

/ M Tokens

Output Price

$

15.0

/ M Tokens

Metadata

Create on

License

Provider

Moonshot AI

HuggingFace

Specification

Architecture

MoE

Calibrated

No

Mixture of Experts

Yes

Total Parameters

2800B

Activated Parameters

Reasoning

Yes

Precision

FP8

Context length

1049K

Max Tokens

1049K

Supported Functionality

Serverless

Supported

Serverless LoRA

Not supported

Fine-tuning

Not supported

Embeddings

Not supported

Rerankers

Not supported

Support image input

Supported

JSON Mode

Supported

Structured Outputs

Not supported

Tools

Supported

Fim Completion

Not supported

Chat Prefix Completion

Not supported

Ready to accelerate your AI development?

Ready to accelerate your AI development?

Ready to accelerate your AI development?