GLM-5.2: The Open Model That Actually Handles Long-Horizon Tasks

Z.ai has released GLM-5.2, a new flagship model specifically tuned for long-horizon work. The key difference from its predecessor is a stable 1-million-token context that doesn't merely "accept" a large input — it actually maintains quality over multi-hour tasks.

The model is especially strong in complex engineering scenarios: from system optimization and writing large codebases to debugging and post-training other models. On the FrontierSWE, PostTrainBench, and SWE-Marathon benchmarks, GLM-5.2 outperforms every open solution and comes very close to top closed models like Claude Opus 4.8.

Another nice touch is the "effort level" control. You can pick a fast mode for routine tasks or switch on the maximum level when top accuracy is required. At the same time, the model remains fully open under the MIT license — no regional restrictions and full access to the weights.

GLM-5.2 is already available on the Z.ai platform and in the repositories on GitHub and Hugging Face. For developers and researchers who regularly work with large code projects, this is one of the most interesting open releases of the past few months.