GLM 5.2 Is Live — Z.ai's Long-Horizon Agent Flagship

June 17, 2026

GLM 5.2 is now in the model picker as Z.ai's upgraded long-horizon agent flagship. It steps up from GLM 5.1 with a 1M-token context window, making it suited for project-level software engineering, long-running agent workflows, and complex multi-step automation that needs to stay in a single reasoning pass.

What you can do

• Pick GLM 5.2 directly from the model selector as GLM 5.2.
• Use a 1M-token context window for large codebases, long briefs, imported document sets, and multi-artifact sessions that would overflow GLM 5.1's window.
• Run it through the SOTA band with reasoning enabled, tool support, and swarm eligibility for harder agentic and engineering workflows.
• Drive long-horizon automation — multi-step plans, project-level software engineering, and tool-heavy orchestration that runs continuously across many turns.
• Keep premium-band pricing in line with GLM 5.1 ($1.40 in / $4.40 out per 1M tokens) while gaining the larger context window.

Where this shows up

• You're working a project-scale engineering task. Where GLM 5.1's ~203K window forced you to trim context, GLM 5.2 holds the whole codebase, design notes, and plan in one pass.
• You're running a long agentic workflow. Pick GLM 5.2 when the session needs to plan, call tools, and keep state across many steps without losing the thread.
• You have a large source set. Drop in long documents, logs, or prior artifacts and use GLM 5.2 when you want a frontier-class long-context window with tool support.

Try it

• "Switch to GLM 5.2 and plan a multi-file refactor of this service, then implement it step by step."
• "Use GLM 5.2 to read all uploaded specs and logs, then produce an implementation plan and a risk list."
• "Use GLM 5.2 with tools to research, build a financial model, and assemble a board deck in one session."

Heads up

• Auto-routable on every plan. GLM 5.2 is in the Auto routing pool for all plans — when Auto escalates to the SOTA tier it can pick GLM 5.2, scored on quality and speed alongside every other model. You can also select it directly from the picker on any plan.
• Premium pricing band. It bills in the premium band, the same as GLM 5.1 — so when Auto chooses it for a hard task it draws from your token budget at the premium multiplier.