Z.ai

Z.ai: GLM Frontier Models for Chat, Coding, and APIs

Z.ai is the AI lab behind the GLM model family, and it sells those models through three surfaces: a free chatbot at chat.z.ai, subscription GLM Coding Plans for agentic coding tools, and a pay-per-token developer API. The lab, which grew out of China's Zhipu AI, also releases many of its models open-weight on Hugging Face and GitHub under the zai-org organization, so the same model line can be rented as a hosted API or downloaded and self-hosted.

You use Z.ai much like OpenAI or Anthropic: chat in the browser for free, plug a Coding Plan into Claude Code, Cline, or another agent harness through its Anthropic-compatible endpoint, or call GLM-5.2, GLM-5-Turbo, and the GLM-4.x line over the API. It is aimed at developers who want frontier-class coding and reasoning models at lower per-token rates, and at teams that want open weights they can run themselves.

Key Features

• GLM-5.2 flagship model built for long-horizon agentic tasks, with a 1M-token context window and published API pricing of $1.40 per million input tokens and $4.40 per million output
• GLM Coding Plan subscriptions in Lite, Pro, and Max tiers with prompt quotas, usable inside Claude Code, Cline, and other coding agents
• OpenAI- and Anthropic-compatible API endpoints, so existing SDKs work with a base-URL change
• Free models for registered users, including GLM-4.7-Flash with a long context window
• Open-weight releases: GLM-4.5 ships as a 355B-parameter mixture-of-experts model with 32B active parameters
• Vision and document models (GLM-5V-Turbo, GLM-4.6V, GLM-OCR) alongside the text line

Ideal Use Case

A developer running an agentic coding tool who wants frontier-level code generation without frontier-level token bills: subscribe to a GLM Coding Plan, point the tool's Anthropic-compatible base URL at Z.ai, and keep the rest of the workflow unchanged. The API tier suits teams building products on GLM models, while the open weights suit infrastructure teams that need to self-host.

How Z.ai differentiates

Z.ai pairs a hosted frontier API with genuinely open releases: GLM-4.5 is published open-weight as a 355B-parameter MoE, the GLM-5 line is documented publicly on GitHub under zai-org, and Z.ai is one of the few providers besides Anthropic to expose an Anthropic-compatible endpoint — which makes GLM a drop-in swap inside Claude Code. Its per-token prices sit well below comparable frontier models from US labs.

FAQ

What is Z.ai? Z.ai is the AI lab (formerly Zhipu AI) behind the GLM model family, offering a free chatbot, coding subscriptions, and a developer API.

Is Z.ai free? The chatbot and several Flash models are free; the GLM Coding Plan and full API are paid, with usage-based or quota-based billing.

Can I use GLM with Claude Code? Yes. The GLM Coding Plan exposes an Anthropic-compatible endpoint, so Claude Code and similar tools can point at GLM models directly.

Are GLM models open source? Many are open-weight — GLM-4.5 and other releases are downloadable from Hugging Face under the zai-org organization.

tl;dr

Z.ai sells the GLM frontier model family through a free chatbot, quota-based coding subscriptions, and a low-priced API, while releasing many of the same models open-weight.