Chinese AI giant Zhipu AI has unveiled GLM-5.2: 1 million token context and open-source code for professionals.
Chinese tech startup Zhipu AI has officially released its new flagship language model GLM-5.2, aimed at solving complex long-term agent tasks and professional programming. This is a fully open-source solution under the MIT license, already available for local deployment. A key feature is a context window of 1 million tokens, opening unprecedented opportunities for analyzing large datasets and working with long code chains.
According to specifications on the Hugging Face platform, GLM-5.2 works with English and Chinese and boasts a colossal 753 billion parameters. However, the real innovation lies in the architecture: the model supports multiple levels of "reasoning intensity," allowing the user to dynamically choose between maximum response accuracy and minimal latency. This makes it a flexible tool for both critical computations and rapid iterations.
At the core of GLM-5.2 are two key architectural innovations: the IndexShare mechanism and an updated MTP layer for speculative decoding. IndexShare significantly optimizes resource consumption by reusing a single indexer for every four layers of sparse attention. Developers claim this reduces operations per token by 2.9 times. In turn, the improved MTP allows increasing the generation confirmation length by up to 20%, directly accelerating output results.
Benchmark results confirm the model's ambitions. In three key tests — FrontierSWE, PostTrainBench, and SWE-Marathon — GLM-5.2 confidently outperformed all other open-source models. In standard programming performance tests, it also took a leading position among open-source solutions, making it a serious competitor to proprietary giants like GPT-4.
GLM-5.2 is distributed under the MIT license, providing maximum freedom for commercial use and modification. For local deployment, support for popular frameworks is announced: SGLang, vLLM, Transformers, KTransformers, and Docker Model Runner. Additionally, quantized versions are available for llama.cpp, Ollama, and LM Studio, allowing the model to run even on hardware with limited resources.
Expert opinion: The release of GLM-5.2 is not just another launch, but a clear signal of a shift in the industry's direction. Chinese developers are no longer catching up and are beginning to set standards, especially in efficiency and openness. The combination of a 1 million token context with open source and the MIT license creates a powerful foundation for the emergence of new, truly decentralized AI agents that do not depend on cloud APIs. This is a step toward democratizing the most advanced technologies.