# The new AI model Z.ai surpassed GPT-5.5 in development tests

Chinese startup [Z.ai](<http://z.ai/>), formerly Zhipu AI, has unveiled a new open-weight AI model, GLM-5.2. It is designed for complex programming, engineering, and large-scale data processing tasks.

The model features 753 billion parameters and supports a context window of up to 1 million tokens. GLM-5.2 is already available via Hugging Face, the Z.ai API, and over 20 development platforms.

According to the company, the new model has demonstrated strong performance in developer-focused benchmarks. On the SWE-bench Pro benchmark, it scored 62.1 points, surpassing GPT-5.5. In tests involving long-horizon technical tasks, GLM-5.2 also approached the performance levels of leading closed-source models.

A key advantage of the model is its cost. Z.ai claims that using GLM-5.2 costs approximately six times less than comparable solutions from OpenAI.

The model is distributed under the MIT license and supports two operational modes: Max for deeper analysis and High for a balance between speed and quality. Developers have already begun integrating GLM-5.2 into programming tools.

Source: [venturebeat.com](<https://venturebeat.com/technology/z-ais-open-weights-glm-5-2-beats-gpt-5-5-on-multiple-long-horizon-coding-benchmarks-for-1-6th-the-cost>)
