GLM-5.2 from Zhipu AI: 1 million token context window and open-source code for agent tasks
Chinese startup Zhipu AI has unveiled its new flagship language model GLM-5.2, designed for long-term agent tasks and complex programming. The open-source solution, distributed under the MIT license, offers a context window of 1 million tokens, dramatically expanding capabilities for working with large documents and code. The model supports local deployment, which is critical for enterprise users with data security requirements.
According to the specification on Hugging Face, GLM-5.2 has 753 billion parameters and generates text in English and Chinese. A key feature is the customizable "reasoning intensity," allowing a balance between response quality and latency. The architecture integrates IndexShare mechanisms and an updated MTP (Multi-Token Prediction) layer for speculative decoding. IndexShare reuses one indexer for every four layers of sparse attention, reducing operations per token by 2.9 times. MTP, in turn, increases confirmation length by up to 20%.
In the FrontierSWE, PostTrainBench, and SWE-Marathon benchmarks, GLM-5.2 surpassed all existing open-source models. In standard programming performance tests, it also took leading positions among competitors. The model is available for local deployment via SGLang, vLLM, Transformers, KTransformers, and Docker Model Runner, and supports quantizations for llama.cpp, Ollama, and LM Studio.
This release is another step by Zhipu AI in strengthening its position in the open AI model market. In my opinion, GLM-5.2 is particularly interesting for developers working with long contexts: 1 million tokens allows processing entire codebases or technical documentation without losing coherence.