GLM-5.3
Zhipu AI (Z.ai) · 2026-08-14 · 753.3B parameters
GLM-5.3 is Z.ai's open-weight post-training update to the GLM-5.2 base model, focused on complex coding and long-horizon tasks. Its reasoning effort supports low, high, and max, with max as the documented default. The released configuration supports 1,048,576 context positions and uses a 78-layer DeepSeek Sparse Attention backbone; compatibility KV-head fields are not treated as conventional GQA heads. Z.ai launched it on August 14, 2026 for all GLM Coding Plan users, with the API marked coming soon; its release notes date the API release to August 18, and the Hugging Face weights followed later in August.