Bottom line: Z.ai, the chat app from Zhipu AI at chat.z.ai, is genuinely free to use for chat, with a limited quota rather than a paid subscription wall. Developers get an even better deal: Z.ai’s official API pricing lists GLM-4.7-Flash and GLM-4.5-Flash at $0 per token. The paid parts are the flagship API models and the flat-rate GLM Coding Plan for heavy coding-agent use.
What is Z.ai?
Z.ai is the international chat platform for the GLM model family. You open chat.z.ai, sign in, and talk to GLM models for writing, questions, document work, and coding help. The platform also offers agent-style tasks such as building websites, creating slides, and analyzing data, in English and Chinese.
What is actually free on Z.ai?
- Free web chat at chat.z.ai with a limited quota — Z.ai does not publish an exact daily message cap, so treat heavy use as metered.
- GLM-4.7-Flash and GLM-4.5-Flash are free per token on the Z.ai API, according to Z.ai’s official pricing documentation.
- Chat and agent features including website building, slide creation, data analysis, and coding assistance.
- English and Chinese interface and conversations.
- Many GLM model weights are published openly for self-hosting, separate from the hosted chat.
What is paywalled?
The flagship hosted API models are pay-per-token — GLM-5-class models are listed at $1.40 per million input tokens and $4.40 per million output tokens in Z.ai’s pricing documentation. If you code with agents all day, Z.ai sells a GLM Coding Plan subscription (Lite, Pro, and Max tiers) that gives a flat allowance inside supported coding tools instead of per-token billing. None of that is needed for ordinary chat.
Pros and cons
Pros
- A real free chat route to a frontier-class model family, not a 7-day trial.
- Free Flash API models make prototyping and small apps cost $0 in model fees.
- Agent features go beyond plain chat: slides, data analysis, and website building.
- Open weights give technical users a self-host escape hatch.
Cons
- The free chat quota is not published as a clear number, so you cannot plan around an exact cap.
- Flagship API use and sustained coding-agent work are paid.
- A younger, fast-growing service — expect occasional rough edges at peak demand.
Who is Z.ai best for?
Z.ai’s free tier is best for everyday users who want a capable second chatbot alongside ChatGPT or Claude, students and writers working in English or Chinese, and developers who want free Flash-model API calls for prototypes, scripts, and small tools before paying for flagship tokens.
FAQ
Is Z.ai really free?
Yes for chat, within a limited quota, and yes for the GLM-4.7-Flash and GLM-4.5-Flash API models, which Z.ai lists at $0 per token. Flagship API models and the Coding Plan are paid.
Do I need a credit card for Z.ai?
No card is needed to start with the free chat or the free Flash API models.
What does the GLM Coding Plan add?
A flat subscription allowance for using GLM models inside supported coding tools, instead of paying per token on the API.
Can I self-host GLM models instead?
Many GLM weights are published openly, so technical users can run them on their own hardware and skip hosted quotas entirely.
Research-based review: free-tier details checked against Z.ai’s published pricing documentation and live coverage in October 2026. No hands-on testing is claimed.