GLM-5: from Vibe Coding to Agentic Engineering

TL;DR

GLM-5 is a cutting-edge foundation model that advances vibe coding to agentic engineering, featuring cost-efficient training, improved autonomy, and superior performance in complex coding tasks.

Contribution

The paper introduces GLM-5, a novel foundation model with DSA-based training, asynchronous reinforcement learning infrastructure, and new RL algorithms for enhanced long-horizon interaction learning.

Findings

Achieves state-of-the-art performance on major benchmarks.

Demonstrates superior capability in real-world coding tasks.

Reduces training and inference costs significantly.

Abstract

We present GLM-5, a next-generation foundation model designed to transition the paradigm of vibe coding to agentic engineering. Building upon the agentic, reasoning, and coding (ARC) capabilities of its predecessor, GLM-5 adopts DSA to significantly reduce training and inference costs while maintaining long-context fidelity. To advance model alignment and autonomy, we implement a new asynchronous reinforcement learning infrastructure that drastically improves post-training efficiency by decoupling generation from training. Furthermore, we propose novel asynchronous agent RL algorithms that further improve RL quality, enabling the model to learn from complex, long-horizon interactions more effectively. Through these innovations, GLM-5 achieves state-of-the-art performance on major open benchmarks. Most critically, GLM-5 demonstrates unprecedented capability in real-world coding tasks, surpassing previous baselines in handling end-to-end software engineering challenges. Code, models, and more information are available at https://github.com/zai-org/GLM-5.