# Zhipu Open-Sources New Flagship Model GLM-5.1

- Category: Flash (快讯)
- Published: 2026-04-08T12:00:00.000Z
- Source date: 2026-04-08
- HTML: https://www.88z.ai/en/news/flash-20260408-d6c4e4c09102361a
- Markdown: https://www.88z.ai/en/news/flash-20260408-d6c4e4c09102361a.md
- Other language: https://www.88z.ai/news/flash-20260408-d6c4e4c09102361a.md

Zhipu has released its open-source flagship model, GLM-5.1, claiming it to be the world's most powerful open-source model. Capable of working independently for over eight hours, it can autonomously handle complex engineering tasks. The model ranks third globally and first among domestic models in coding benchmarks like SWE-Bench Pro, surpassing GPT-5.4 and Claude Opus 4.6. Tests demonstrate its ability to build complete Linux desktop systems and significantly optimize vector database and ML work

## At a glance

- GLM-5.1 ranks third globally and first domestically in coding, surpassing GPT-5.4 and Claude Opus 4.6
- Supports over 8 hours of independent work, capable of building Linux systems and significantly optimizing database and ML workloads

## The update

Zhipu has officially open-sourced its new flagship model, GLM-5.1, positioning it as the current most powerful open-source model globally. This model features strong autonomous agent capabilities, allowing it to work independently for over eight hours to complete various complex engineering tasks. In terms of coding proficiency, GLM-5.1 performs excellently in authoritative benchmarks such as SWE-Bench Pro, ranking third worldwide and first among domestic models, with performance metrics surpassing GPT-5.4 and Claude Opus 4.6. Real-world application tests show that the model can build a complete Linux desktop system within eight hours, optimize vector database performance by nearly seven times, and achieve a 3.6x acceleration in ML workloads through twenty-four hours of iterative optimization.

## Sources

- [智谱](https://mp.weixin.qq.com/s/F44gMXoPZhILc5nEoJ9p_A) (2026-04-08)

This article is an AI-assisted synthesis by 88Z AI based on the public sources below. Product claims and availability are attributed to their publishers. It does not represent independent testing or first-hand reporting.