KAT-Coder-V2 Technical Report

TL;DR

KAT-Coder-V2 is a modular agentic coding model that combines specialized expert domains through supervised fine-tuning and reinforcement learning, achieving state-of-the-art results across multiple benchmarks.

Contribution

It introduces a

Findings

Achieves 79.6% on SWE-bench Verified, close to Claude Opus 4.6's 80.8%.

Surpasses GLM-5 and MiniMax M2.7 on PinchBench.

Ranks first in frontend aesthetics scenarios.

Abstract

We present KAT-Coder-V2, an agentic coding model developed by the KwaiKAT team at Kuaishou. KAT-Coder-V2 adopts a "Specialize-then-Unify" paradigm that decomposes agentic coding into five expert domains - SWE, WebCoding, Terminal, WebSearch, and General - each undergoing independent supervised fine-tuning and reinforcement learning, before being consolidated into a single model via on-policy distillation. We develop KwaiEnv, a modular infrastructure sustaining tens of thousands of concurrent sandbox instances, and scale RL training along task complexity, intent alignment, and scaffold generalization. We further propose MCLA for stabilizing MoE RL training and Tree Training for eliminating redundant computation over tree-structured trajectories with up to 6.2x speedup. KAT-Coder-V2 achieves 79.6% on SWE-bench Verified (vs. Claude Opus 4.6 at 80.8%), 88.7 on PinchBench (surpassing GLM-5 and MiniMax M2.7), ranks first across all three frontend aesthetics scenarios, and maintains strong generalist scores on Terminal-Bench Hard (46.8) and tau^2-Bench (93.9). Our model is publicly available at https://streamlake.com/product/kat-coder.