Z.AI

Z.ai's June 2026 open-weight (MIT) model for long-horizon coding, the first GLM to deliver a usable 1M-token context, with 128K output and selectable reasoning effort. Superseded as flagship by GLM-5.3, a post-trained update of the same base.

Z.ai's February 2026 open-weight (MIT) foundation model for agentic engineering: a 744B MoE with 40B active parameters, 200K context and optional thinking. Superseded by GLM-5.1, GLM-5.2 and now GLM-5.3.

Z.ai's main image model since January 2026, pairing a 9B autoregressive generator with a 7B diffusion decoder. Open-weight (MIT) and aimed at accurate in-image text for posters, slides and diagrams, at 512 to 2048 px.

Lightweight high-speed edition of GLM-4.7, the paid counterpart to the free GLM-4.7-Flash, with a 200K context and 128K output for coding and agent tasks. Z.ai's newer fast tier is GLM-5.3-FlashX.

Z.ai's open-weight (MIT) coding-focused model from December 2025, with a 200K context, 128K output and reasoning that can be kept across turns and tool calls. Superseded by the GLM-5 series, now led by GLM-5.3.

Z.ai's first open-source text-to-image model (6B, Apache 2.0, March 2025), notable for drawing Chinese characters in images and taking long Chinese or English prompts; outputs 512 to 2048 px. GLM-Image has since replaced it as Z.ai's main image model.