KAT-V1-40B

Kwaipilot· 20 Jul 2025

General purpose

A hybrid reasoning model which was trained with Step-SRPO to reward the model selecting the correct reasoning mode.

Specs

Params40B

LicenseKwaipilot License

Tags

modelsmodels/llmsmodels/llms/reasoningmultilingualartifacts/13

Resources

Adoption · Interconnects

RAM score

—

Hugging Face Downloads

last 30d

4.8K

all time

HF Likes

118

Relative Adoption Metric contextualizes downloads against the model's size bucket.

VAIL Fingerprint

0102:016a:0218:049a:093d:0fe0:1e1a:590e

Explore other models with behavioral similarity to KAT-V1-40B.

Related Models

More from Kwaipilot