Multimodal model

Kimi K2.7 Code

Kimi K2.7 Code is a coding-focused model built on a 1T parameter MoE architecture with 32B active parameters and a 256K context window. It features native multimodal input (text, image, video), mandatory long-thinking mode, tool calling, and enhanced agentic coding. It reduces overthinking by 30% vs K2.6 and achieves a 21.8% improvement on Kimi Code Bench v2.

Type
Multimodal
Weights
Open
Downloads
97k
Likes
1.4k
Pipeline
image text to text
License
other

Model hub

moonshotai/Kimi-K2.7-Code

Repo
moonshotai/Kimi-K2.7-Code
Library
transformers
Access
Public
Technical tags and languages
transformers safetensors kimi_k25 feature-extraction compressed-tensors image-text-to-text conversational custom_code eval-results
Open on Hugging Face

Related tools

5

More from Moonshot AI

10

Pricing

2

About Moonshot AI

1

Moonshot AI is a Beijing-based Chinese AI startup (founded 2023) focused on developing large language models and multimodal systems

Benchmark scores

7

How Kimi K2.7 Code ranks on tracked AI benchmarks. Click any benchmark to see its full leaderboard.

Tools using Kimi K2.7 Code

40