---
title: "KNOWLEDGE ATLAS released the GLM-5 technical report, with all technical details disclosed"
description: "KNOWLEDGE ATLAS released the GLM-5 technical report, introducing its next-generation foundational model GLM-5, which aims to shift the programming paradigm from \"VibeCoding\" to \"AgenticEngineering.\" G"
type: "news"
locale: "en"
url: "https://longbridge.com/en/news/276532606.md"
published_at: "2026-02-22T10:20:26.000Z"
---

# KNOWLEDGE ATLAS released the GLM-5 technical report, with all technical details disclosed

> KNOWLEDGE ATLAS released the GLM-5 technical report, introducing its next-generation foundational model GLM-5, which aims to shift the programming paradigm from "VibeCoding" to "AgenticEngineering." GLM-5, based on its predecessor model, employs sparse attention technology to reduce inference costs while maintaining long-context capabilities. The new asynchronous reinforcement learning infrastructure enhances training efficiency, and the new asynchronous Agent reinforcement learning algorithm improves the model's learning ability in complex interactions. GLM-5 performs exceptionally well in mainstream open benchmark tests, particularly surpassing all open-source baselines in real-world programming tasks

KNOWLEDGE ATLAS states that GLM-5 is a next-generation foundational model aimed at shifting the programming paradigm from "VibeCoding" to "Agentic Engineering." Building on the Agentic, Reasoning and Coding (ARC) capabilities of its predecessor GLM-4.5, GLM-5 employs DeepSeek Sparse Attention (DSA) to significantly reduce reasoning costs while maintaining long-context capabilities without loss. To better align the model with various tasks, a new asynchronous reinforcement learning (RL) infrastructure has been constructed, decoupling the generation process from the training process, thereby greatly enhancing post-training iteration efficiency. Additionally, a brand-new asynchronous Agent reinforcement learning algorithm has been proposed, further improving the effectiveness of reinforcement learning, allowing the model to learn more effectively from complex, long-range interactions. Based on these innovations, GLM-5 has achieved SOTA performance in mainstream open benchmark tests. Most importantly, GLM-5 has demonstrated unprecedented capabilities in real-world programming tasks, surpassing all previous open-source baselines in handling end-to-end software engineering challenges

### Related Stocks

- [02513.HK - KNOWLEDGE ATLAS](https://longbridge.com/en/quote/02513.HK.md)

## Related News & Research

| Title | Description | URL |
|-------|-------------|-----|
| 中金：予智谱 “跑赢行业” 评级 目标价 688 港元 | 中金发布研报，给予智谱 “跑赢行业” 评级，目标价 688 港元。预计 2025-2028 年收入为 7.2/17.5/36.7/76.7 亿元，CAGR 为 120%。公司专注 GLM 系列基座模型，最新一代 GLM-5 在多个基准评分中 | [Link](https://longbridge.com/en/news/276405013.md) |
| 下沉市场引爆 AI 应用需求，汇添富港股通科技精选混合发起式 C(025545) 精准布局港股科技，把握成长机遇 | 下沉市场推动 AI 应用需求激增，春节期间用户规模大幅增长。科技巨头投资超 45 亿元抢占市场，AI 日活跃用户从 1-2 亿增至 3-4 亿。港股 AI 概念股集体上涨，智谱涨 42.72%，海致科技涨超 28%。汇添富港股通科技精选混合 | [Link](https://longbridge.com/en/news/276505731.md) |
| 无视海外 “AI 恐慌”，中国市场狂炒 AI 赢家 | 同一个 AI，截然相反的命运。美国市场陷入” 恐慌性抛售”，中国投资者疯狂追捧：智谱 AI 一度暴涨 524%，MiniMax 一度飙升 488%。分歧本质是美国投资者焦虑于丰厚利润池面临竞争威胁，而中国关注的仍是市场渗透。 | [Link](https://longbridge.com/en/news/276531940.md) |
| 智谱上市 1 月涨 5 倍，市值超越京东、快手 | 智谱 AI 股价在上市 43 天内累计涨幅超过 500%，市值突破 3232 亿港元，超越京东和快手，成为港股 TMT 板块市值第一梯队。智谱于 2 月 12 日发布新一代大模型 GLM-5，参数总量达 744B，市场反响热烈，导致其产品价 | [Link](https://longbridge.com/en/news/276435403.md) |
| 理解市场 \| 智谱高开逾 5% 中科曙光旗下 SothisAI 接入智谱 GLM-5 | 智谱高开逾 5%，截至发稿，涨 5.12%，报 534 港元，成交额 2461.74 万港元。消息面上，据中科曙光消息，曙光 SothisAI 近日升级，实现智谱 GLM-5、阿里 Qwen3.5 大模型全面接入，并进一步简化和优化企业 A | [Link](https://longbridge.com/en/news/276400302.md) |

---

> **Disclaimer**: This article is for reference only and does not constitute any investment advice.