Listing checkedFreemiumZ.ai
Zhipu AI's international chat and agent platform on its GLM models, with free web chat, credit-metered coding plans for Claude Code, Cline and OpenCode, and a per-token API.
- Ratings
- No member reviews yet
- Price
- From $18/mo
- Last updated
Key takeaways
- Z.ai is the international chat, agent and developer platform of Zhipu AI, the Beijing company behind the GLM model family. As of August 2026 the homepage bills it as an AI chatbot and agent powered by GLM-5.3-Flash, and the same GLM models are sold through paid coding subscriptions and a per-token API. It is a model vendor's front door, not a wrapper around other people's models.
- The focus has shifted to developers. The scrape still describes a free assistant for presentations, writing and coding, but Zhipu's current documentation leads with GLM Coding Plans, credit formulas and integrations for Claude Code, Cline and OpenCode, while the consumer chat keeps the broader pitch.
- Web chat is free and coding plans are not. As of August 2026 the GLM Coding Plans cost $18, $80 and $168 per month for Lite, Pro and Max, metered in rolling five-hour and weekly credit allowances since a July 30, 2026 revision. Subscriptions auto-renew and purchased terms are non-refundable, unused or not.
- Traffic is large and climbing. A third-party estimate puts chat.z.ai at roughly 10.4M visits for June 2026, up from about 4.2M in November 2025, with around 73% arriving direct. The figure was captured in August 2026 and has not been independently confirmed.
- The biggest caveat is hosted reliability. Independent reviews and Trustpilot posts from February and March 2026 describe multi-day API outages, throttling, lost chat history, unanswered support tickets and bans after complaints, and DigiTimes reported that Z.ai apologised on February 21, 2026 after a GLM-5 rollout drew backlash over billing and throttling. The model itself gets praised. The service around it doesn't.
What is Z.ai?
Z.ai is the international brand of Zhipu AI, the Beijing company formally named Beijing Zhipu Huazhang Technology, and it puts Zhipu's GLM models in front of a global audience through three doors. The first is a browser chat and agent at chat.z.ai, which is free to use and, as of August 2026, runs on GLM-5.3-Flash. The second is a set of paid GLM Coding Plans that route the models into coding tools such as Claude Code, Cline and OpenCode. The third is a per-token API for developers who want GLM-5.3, its Flash variant or the older GLM-4.x models inside their own software.
The models are Zhipu's own, and that is what sets it apart. GLM-5.3, GLM-5.2, GLM-5.1, GLM-5, the GLM-4.7 and GLM-4.5 variants and GLM-4.6 all appear in the current documentation, and several GLM releases have shipped as open weights under open licences. The hosted service and the open-weight licence are not the same thing, though, and the terms for each need reading separately. The documentation also lists three tool services, Web Search, Web Reader and Zread, each metered in credits, plus vision support in the developer platform.
The scrape and third-party listings still lead with AI Slides, presentation generation and downloadable structured outputs. An official Slides feature page could not be retrieved in August 2026, so treat that part of the pitch as unconfirmed until you see it in the product. The confirmed parts are a chat and agent product, a coding subscription business and an API, all priced far below the mainstream Western assistants.
How it works
Open the chat
Go to chat.z.ai in a browser and type a request. Free consumer access is advertised, though the exact quota and whether an account is required were not confirmed in August 2026. The homepage identifies GLM-5.3-Flash as the model behind the chat.
Give it context and a job
Paste or upload the material you want it to work from where the interface allows, then ask for writing, code, a summary, a research pass or a longer agent task. Third-party descriptions mention document analysis and structured downloadable outputs, but those details need checking against the live interface.
Route the coding work
For coding, choose a GLM Coding Plan and connect a supported tool: the documentation names Claude Code, Cline and OpenCode. Every request then draws down credits from a five-hour window and a weekly allowance, with the amount depending on the model, the tokens in and out, and any Web Search, Web Reader, Zread or MCP calls.
Review and export
Read the answer, iterate and copy or export the result. The exact export formats for presentations were not verified in primary documentation, so check the interface before promising a deck to anyone.
Features
GLM Chat and Agent
The consumer product is a general chat and agent interface for conversation, reasoning, writing and longer multi-step tasks, running on the GLM family. As of August 2026 the homepage names GLM-5.3-Flash as the model in use.
GLM Coding Plan
A subscription for coding tools and heavier usage, sold in Lite, Pro and Max tiers since the July 30, 2026 revision. Each tier carries a five-hour credit window and a weekly cap, and off-peak usage is charged at half the standard credit rate.
Code generation and explanation
Z.ai generates, debugs, explains and refactors code, and third-party descriptions single out Python. Through the coding plans the same models sit behind agentic tools rather than a chat box.
Web Search, Web Reader and Zread
Three documented tool services, exposed as MCP components, that fetch search results, read web pages and, in Zread's case, apparently cover reading and research workflows. Each has its own credit multiplier, so a research-heavy session spends faster than a plain chat.
Developer integrations
The developer overview lists compatibility with Claude Code, Cline and OpenCode, and describes vision support. This is where Zhipu's documentation puts most of its current attention.
API model access
GLM models are available through a separately priced token API. As of August 2026 the table lists GLM-5.3 at $1.40 per million input tokens, $0.26 for cached input and $4.40 for output, GLM-5.3-Flash on promotional rates until September 9, 2026, and GLM-4.7-Flash and GLM-4.5-Flash as free.
AI Slides
The scrape and third-party listings claim presentation generation from a prompt. No current official feature page for Slides was retrieved in August 2026, so its availability and export formats should be confirmed in the product before relying on it.
What people use it for
- Drafting emails, reports, research text and marketing copy.
- Generating, explaining, debugging and refactoring Python and other code.
- Running GLM models inside Claude Code, Cline or OpenCode on a coding plan.
- Searching and synthesising web information through Web Search and Web Reader.
- Summarising long documents and text.
- Brainstorming projects, content and presentation ideas.
- Building slide or deck outlines, where AI Slides is enabled.
- Automating repetitive developer or research tasks over the API.
It is not a dependable substitute for expert review of factual research, production code or confidential material. The terms permit Zhipu to use non-personal user content for improvement, and users reported outages, throttling and lost chat history through early 2026.
Who it’s for
Pricing
Official pricingIndividual plans
Free
$0
per month
Browser chat on GLM-5.3-Flash; quota and login rules not confirmed as of August 2026
Lite
$18
per month
GLM Coding Plan: 2,000 credits per five-hour window and 10,000 per week; $12.60 effective monthly on annual billing
Pro
$80
per month
GLM Coding Plan: 12,000 credits per five-hour window and 60,000 per week; $56 effective monthly on annual billing
Max
$168
per month
GLM Coding Plan: 28,000 credits per five-hour window and 140,000 per week; $117.60 effective monthly on annual billing
Web chat is free, and as of August 31, 2026 the paid product is the GLM Coding Plan at $18 per month for Lite, $80 for Pro and $168 for Max, with annual billing shown as effective monthly rates of $12.60, $56 and $117.60; the pages don't say clearly whether the annual price is charged upfront. Since July 30, 2026 the plans are metered in credits, not prompts: Lite gets 2,000 credits per five-hour window and 10,000 per week, Pro 12,000 and 60,000, Max 28,000 and 140,000. A credit isn't one generation. Usage is calculated from input, cached-input and output tokens with model-specific multipliers, GLM-5.3 anchors the formula, and Web Search, Web Reader, Zread and MCP calls all draw credits too. Five-hour credits refresh five hours after they were consumed and weekly credits reset every seven days. Weekday peak hours are 14:00 to 18:00 Singapore time, and off-peak model usage is charged at 50% of the standard rate. API use is billed separately per million tokens: GLM-5.3 at $1.40 input, $0.26 cached input and $4.40 output, GLM-5.3-Flash at a promotional $0.075, $0.015 and $0.25 until September 9, 2026 at 24:00 UTC+8, and GLM-4.7-Flash and GLM-4.5-Flash free. No trial was found. Paid subscriptions renew automatically, and the two official pages disagree on notice: the subscription terms say cancel at least 24 hours before the term ends, a developer help page says three days before billing. Cancelling keeps access until the paid term ends, and refunds are not supported, including for unused subscription value. No team, business or enterprise price table was found as of August 2026, and seats, commercial rights, SSO and pooled credits are not documented.
Traffic & reach
Third-party estimates from an August 2026 capture, not official numbers, and in this case unconfirmed by any Similarweb, Cloudflare or company source. Taken at face value, about 73% of visits arrive direct and the average session runs over 18 minutes. That is the pattern of a bookmarked daily tool rather than a search find. The audience is spread across China, India, the United States, Russia and Indonesia, none of them above 12% of visits.
Visits, June 2026
10.4M
▲ 23% month over month
Nov '25Jun '26
Top countries
- China12%
- India11%
- United States9.5%
- Russia7.0%
- Indonesia5.0%
Traffic sources
- Direct73%
- Organic search14%
- Referrals6.6%
- AI assistants2.8%
- Organic social2.2%
- Visits, last 3 months
- 27.4M
- Avg. visit
- 18m 13s
- Pages per visit
- 3.4
- Bounce rate
- 23.7%
- Brand searches / mo
- 2.5M
User reviews
Write a reviewNo reviews yet
Used Z.ai on something real? Write down what you made, what it cost and what the trade-off was. The next person can decide in about a minute from that.
Be the first to reviewWhat to know before you commit
The draw is cheap GLM access. Reviewers and users describe Z.ai as an unusually inexpensive route to GLM chat and coding, with a free consumer entry point, multiple coding-tool integrations and bilingual access in one place. The GLM models themselves are often praised, including by users who otherwise criticise the hosted service; one February 2026 Trustpilot reviewer called GLM 5 an excellent model whose only downside was the human factor.
Reliability and support are the complaint pattern. A March 2026 independent review, which drew on a Reddit thread and Discord reports, reported multi-day API outages, no status-page updates and unanswered support tickets, and named API unreliability as the most common user complaint. Trustpilot posts from February and March 2026 allege throttling, wiped chat history, bans after complaining on Discord, weak coding quality and a gap between advertised and delivered usage. These are user allegations, not proven findings, but they are consistent enough to weigh before paying.
The GLM-5 rollout ended in an apology. DigiTimes reported that Z.ai issued a public apology on February 21, 2026 after the GLM-5 rollout drew backlash over billing, throttling and flaws in the upgrade mechanism. DigiTimes framed this as an operational and product problem, not a confirmed security breach, and no independently verified data breach or lawsuit was found.
Your content can be used for improvement. The privacy policy says prompts, images and other user content are processed to provide the service, and the terms reserve the right, for individual users, to use and store non-personal user content to improve the service and its AI, including processing outside your jurisdiction. Don't assume a no-training default, and read the current terms if you handle confidential material.
The numbers around it are unverified. No aggregate G2, Capterra, Product Hunt or app-store rating with a review count could be retrieved, so the reception picture is built from anecdotes rather than a representative sample. Financing figures come from secondary reporting that conflicts on later valuation, and the traffic estimate has no independent corroboration.
Alternatives to Z.ai
ChatGPTOpenAI's conversational AI assistant for drafting, coding, research and data analysis, with a permanent free tier and paid plans on web, iOS and Android.
CrushOn.AIAdult character roleplay and AI companion chat with user-created characters, plan-based memory up to 24K and a permanent free tier on free models.
GeminiGoogle's AI assistant for writing, research, coding and everyday questions, free with a Google Account and upgradable through Google AI Plus, Pro and Ultra plans.
GrammarlyAI writing assistant that checks grammar, clarity and tone and drafts text across the browser, Office, desktop and mobile, with a permanent free plan.
ElevenLabsText to speech, voice cloning, dubbing and voice agents on one credit balance, in a browser studio and behind a REST API.
LovableDescribe a web app in chat and Lovable writes real full-stack code, provisions the database, auth and hosting, and syncs the repo to GitHub or GitLab.
Frequently asked questions
Company & links
- Company
- Z.ai
- Founded
- 2019
- Founders
- Tang Jie, Li Juanzi
- Leadership
- Available reporting names Zhang Peng as CEO and Liu Debing as chairman, though the current executive roster was not confirmed from primary filings. Tang Jie, a Tsinghua University computer-science professor, is identified as founder alongside Li Juanzi; the company grew out of Tsinghua research into language models and knowledge graphs. The exact formal titles of each founder remain an editorial check.
- Funding
- Secondary reporting lists roughly $137 million in March 2024, about $400 million in June 2024, about $411 million in December 2024 and a reported $140 million round in July 2025, with a total near $1.5 billion since founding, though the rounds may overlap and no primary ledger was retrieved. Secondary reporting says the company listed in Hong Kong in January 2026 as Knowledge Atlas Technology, raising about $558 million at a valuation around $6.5 billion; secondary sources conflict on later valuation, so check the exchange filing. Named investors include Alibaba, Ant Group, Tencent, Meituan, Xiaomi, HongShan, Saudi Aramco and Prosperity7 Ventures.
- Scale
- A retrieved reference gives 800 or more employees in 2024, with no current headcount verified. No reliable self-reported revenue or profitability figure was found. The company is described as public after the Hong Kong listing and has not been acquired.
- Contact
- user_feedback@z.ai
Try Z.ai on work you already know
A task you’ve done before is the fastest way to judge whether Z.ai fits your workflow and budget. Check the current plan terms on the official site before you subscribe.