Z.ai
: How to use it, features, and the business problems it solves
Add bookmark
What is Z.ai?
The official service for the GLM family of large language models developed by Z.ai (Zhipu AI). In chat (chat.z.ai), GLM-5.3-Flash can write, analyze data and build slides and websites. For developers, Z.ai offers an API (GLM-5.3 costs $1.4 input / $4.4 output per 1M tokens) and the "GLM Coding Plan", a flat-rate plan for using GLM in 20+ tools such as Claude Code and Cline (Lite $18, Pro $80, Max $168 per month). Model weights are published on Hugging Face. Officially supported model languages are English and Chinese, and the interface is in English.
Business problems it solves
About "Z.ai"
What is Z.ai?
Z.ai is the official service of Z.ai, the developer of the GLM family of large language models (its official Hugging Face organization page reads "Zhipu AI (Z.ai)"). It offers three things: the browser-based AI chat "chat.z.ai", an API for developers, and the "GLM Coding Plan", a flat-rate plan for using GLM in coding agents.
As of October 2026, the latest models are the flagship "GLM-5.3" (released August 18) and the multimodal "GLM-5.3-Flash" (released August 26). Both handle a 1M-token context, and the weights of GLM-5.3-Flash are published on Hugging Face. GLM models are also available on Amazon Bedrock.
Chat that goes from writing to slides and websites
chat.z.ai is the AI chat and agent interface powered by GLM-5.3-Flash. The chat screen shows examples labeled "Full-Stack", "Writing", "Data Insight", "PPT" and "Web Site", so you can ask for data analysis, slide decks and websites, not just text. You can pick the model (such as GLM-5.3-Flash) at the bottom right of the input box.
Source: Z.ai official website (the images below are from the GLM Coding Plan page)
GLM-5.3-Flash is a multimodal model that can read images, video and files. According to the official documentation, beyond coding it supports office documents and financial research, and can deliver finished PPTX, PDF, DOCX and XLSX files.
How your input is handled: The chat's terms of use state that content entered by individual users may be used to improve the service and develop new services. The Team plan page, on the other hand, states that "your data is never used for model training." If you plan to enter confidential business information, check which contract you are on first.
GLM Coding Plan: flat-rate GLM for coding agents
The GLM Coding Plan lets you use GLM at a fixed monthly price in more than 20 agent tools, including Claude Code, Codex, Cline, OpenCode, Cursor and Kilo Code. Instead of paying per API call, you consume credits within the plan's allowance. It also works with the official IDE "ZCode" and the work-automation tool "AutoClaw".
- Models: GLM-5.3 and GLM-5.3-Flash on all plans. Requests for GLM-5.2/GLM-5.1 are routed to GLM-5.3, and requests for GLM-4.7 to GLM-5.3-Flash
- Usage limits: both a 5-hour limit and a weekly limit apply
- Half price off-peak: outside weekdays 14:00–18:00 Singapore time (15:00–19:00 Japan time), credits are consumed at 50% of the standard rate
- Bundled MCP tools: Vision Understanding, Web Search, Web Reader and Zread are available on all plans
Only inside supported tools: The usage policy says the plan may only be used in officially supported tools, and use elsewhere may result in restricted benefits. Account sharing is prohibited. You cannot call the API directly with the plan's quota; API usage is billed separately on a pay-as-you-go basis.
How to use
- Create an account at chat.z.ai Open chat.z.ai in your browser, sign in and start chatting.
- Ask for what you want to make Describe the text, document or website you want in the input box. The examples on the screen (Writing, Data Insight, PPT, Web Site, etc.) are a good starting point. You can attach files and images.
- For development, issue an API key Create an API key on the Z.ai API platform and call the API with a model name such as glm-5.3 or glm-5.3-flash.
- For coding tools, subscribe to the GLM Coding Plan Choose Lite, Pro or Max on the pricing page, then configure tools such as Claude Code by following the official quick start.
Model highlights
01GLM-5.3 (flagship)
- Stronger coding — Z.ai reports a 50% gain over GLM-5.2 on its own Z.ai Code Bench and state-of-the-art results among open-source models on public benchmarks
- Long tasks — designed to take a single task from requirements to deployable artifacts within a 1M-token context
- Security — Z.ai also highlights white-box code review and vulnerability discovery
02GLM-5.3-Flash (multimodal)
- Low cost and speed — 320B total parameters with 18B active. Combines sparse and linear attention to cut the cost of long-context serving
- Builds while looking at the screen — observes interfaces, rendered results and interaction feedback, working across code, browsers and GUIs
- Faster variant — GLM-5.3-FlashX responds at 200 tokens/s via the API (not yet on the Coding Plan)
03Other API features
- Developer features — thinking mode, streaming, function calling, context caching, structured output
- Built-in tools and agents — web search, image generation (GLM-Image), a slide/poster agent, and a translation agent supporting 40+ languages
- Free models — GLM-4.7-Flash, GLM-4.5-Flash and GLM-4.6V-Flash are free on the API
Pricing
GLM Coding Plan (individual)
Monthly price per user (official pricing page as of October 2026, in USD). Quarterly billing is 20% off and yearly billing is 30% off.
| Plan | Monthly | Quarterly (per month) | Yearly (per month) | Allowance |
|---|---|---|---|---|
| Lite | $18 | $14.4 | $12.6 | 10,000 credits/week (2,000 per 5 hours) |
| Pro | $80 | $64 | $56 | 6× Lite (60,000/week, 12,000 per 5 hours) |
| Max | $168 | $134.4 | $117.6 | 14× Lite (140,000/week, 28,000 per 5 hours) |
What the credits mean: The official documentation estimates that at a 95% cache hit rate, Lite gives 48–97 million GLM-5.3 tokens per week (the lower bound if all usage is at peak, the upper bound if all is off-peak). A Team plan is also available, with centralized seat and permission management and a usage dashboard. Subscriptions renew automatically and are non-refundable once purchased.
API (pay-as-you-go)
Price per 1M tokens (official pricing page as of October 2026, in USD).
| Model | Input | Cached input | Output |
|---|---|---|---|
| GLM-5.3 | $1.4 | $0.26 | $4.4 |
| GLM-5.3-Flash | $0.15 | $0.03 | $0.5 |
| GLM-5.3-FlashX | $0.37 | $0.075 | $1.25 |
| GLM-4.7-Flash / GLM-4.5-Flash | Free | Free | Free |
Web search costs $0.01 per use, and GLM-Image costs $0.015 per image.
Pros and cons
Pros
- Use GLM at a flat rate from $18/month inside familiar coding tools such as Claude Code
- Off-peak usage (outside weekday 15:00–19:00 Japan time, and all weekend) consumes half the credits
- Low API prices: GLM-5.3-Flash is $0.15 input and $0.5 output per 1M tokens
- GLM-5.3-Flash weights are on Hugging Face, so self-hosting is an option
Cons
- The site and interface are in English, and the officially supported model languages are English and Chinese. You need to test how well it works in Japanese yourself
- The terms state that individual users' input may be used to improve the service
- The GLM Coding Plan is non-refundable and cannot be used outside supported tools
How Z.ai differs from other Chinese AI models
| Product | Provider | Highlights | Pricing (as listed on each product page at the time of checking) |
|---|---|---|---|
| Z.ai | Z.ai (Zhipu AI) | GLM-5.3 / 5.3-Flash. Centered on a flat-rate plan for coding tools | Coding Plan from $18/month / pay-as-you-go API |
| DeepSeek | DeepSeek | Weights published under the MIT license. Pay-as-you-go API at half price off-peak | Free chat / pay-as-you-go API |
| Qwen | Alibaba | Image and video generation and web search in chat, free | Free chat |
| Kimi | Moonshot AI | Agents for slides, Deep Research and more. Web version has a Japanese interface | Free plan; Plus from $19/month |
If you want a flat monthly price rather than pay-as-you-go for the model behind your coding tools, Z.ai's GLM Coding Plan is the one to compare. For Anthropic's own agent, see Claude Code; for open-source agents where you can swap models, combine it with Cline or Kilo Code. If you want a chat interface in Japanese, Kimi or Qwen is easier to start with.
FAQ
Q. What is the relationship between Z.ai and GLM? A. GLM is the name of the large language model family developed by Z.ai. Z.ai is the name of the service (and the company) that provides GLM via chat, API and a flat-rate coding plan.
Q. How much does it cost? A. The GLM Coding Plan is $18/month for Lite, $80/month for Pro and $168/month for Max (30% off with yearly billing). The API is pay-as-you-go; GLM-5.3 costs $1.4 input and $4.4 output per 1M tokens.
Q. Which tools work with the GLM Coding Plan? A. More than 20 tools, including Claude Code, Codex, Cline, OpenCode, Cursor, Kilo Code and the official IDE ZCode. It cannot be used in tools that are not officially supported.
Q. Can I use it in Japanese? A. The official documentation lists English and Chinese as GLM's supported languages; Japanese is not named. The site and chat interface are in English as well. If you plan to use it in Japanese, try the chat first before deciding.
Q. Where is my data processed? A. According to the privacy policy, the service is generally provided from Singapore and personal data is generally processed in Singapore.
Q. Can I run the model myself? A. The weights of GLM-5.3-Flash are published on Hugging Face, and the official guide says it runs on inference frameworks such as SGLang and vLLM.
The information on this page is based on what was published on Z.ai's official website and documentation as of October 2026. Please check the official website for the latest specifications and pricing.
