AutoGLM (Zhipu)
Zhipu's autonomous agent that controls Android phones
Overview
AutoGLM is Zhipu AI's autonomous-agent product that controls real Android phones to complete multi-step tasks — book a ride, order food, send WeChat messages, search and act. Pre-installed on selected Chinese Android phone models in partnership with handset OEMs.
---
AutoGLM 是智谱 AI 推出的自主智能体产品，可控制真实 Android 手机完成多步任务——叫车、点餐、发微信消息、搜索并操作。已与多家国产手机厂商合作，预装于特定机型。
Pricing
Pricing shown for reference only. These figures reflect RECATOOLS research as of 20 May 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.
Use cases
What you can produce with AutoGLM (Zhipu)
- Tell AutoGLM to order a specific meal on Meituan and have it open the app, search the restaurant, assemble the cart, and stop at the payment confirmation step.
- Ask the agent to book a ride-hailing car to a stated destination, with the agent handling app navigation, pickup-point selection, and ride-type choice on your behalf.
- Dictate a WeChat message to a named contact and have the agent find the conversation, type the message, and send it without you touching the screen.
- Chain a multi-app errand, such as searching a product on Taobao, comparing a few listings, and adding the chosen one to the cart, in a single natural-language instruction.
- Run tasks that span more than fifty individual UI steps, letting the agent recover from intermediate screens like pop-ups and login prompts as it goes.
- Download the Open-AutoGLM framework and the AutoGLM-Phone-9B model to build and self-host your own Android phone agent on top of Zhipu's open-source stack.
- Use AutoGLM pre-installed on partner Chinese Android handsets as a system-level assistant rather than installing a separate app.
ASEAN Perspective
AutoGLM (Zhipu) in Southeast Asia
与海外 Operator、Claude Computer Use 形成对照，差异在于 AutoGLM 跳过浏览器直接操作 Android 应用界面。
AutoGLM is Zhipu AI's execution-focused agent that operates phones, desktops, and web by clicking, typing, and swiping to complete real tasks, with v2 adding cloud-phone execution so jobs run without tying up your device. It is technically ambitious GUI-agent work and competitively cheap per task.
It suits users embedded in the Chinese app ecosystem, since its core workflows target apps like WeChat, Taobao, Meituan, and Douyin; outside China its practical utility drops sharply. For ASEAN users this is the major caveat: app coverage, language, and account requirements are China-centric, and data flows through a Chinese provider, which raises governance and PDPA considerations. Best treated as a capability preview for most regional buyers.
What people say
AutoGLM is Zhipu AI's (now branded Z.ai) phone-use agent, and it is one of the few agent products that actually operates a real Android device rather than a browser sandbox. Launched into public beta in late 2024 and formally released in 2025, it can plan and execute multi-step tasks across more than fifty high-frequency Chinese apps, including WeChat, Taobao, Meituan, and Douyin: ordering food, booking rides, sending messages, and searching then acting on results. Zhipu has partnered with Chinese handset makers to pre-install it on selected models, and upgrades extended it to task chains exceeding fifty steps across multiple apps.
Because AutoGLM lives almost entirely in the Chinese app ecosystem, there is no Western review-site footprint, no G2 or app-store consensus an English-speaking reader can lean on. What does exist are hands-on write-ups and developer activity around Open-AutoGLM, the open-source framework and AutoGLM-Phone-9B model Zhipu released in December 2025, which drew real attention on GitHub and Hugging Face. Zhipu's own benchmark claims, such as strong results against Western agents on device-use tests, are self-reported and should be read that way.
Independent hands-on testing paints a consistent picture: AutoGLM is genuinely useful for short, repeatable, low-stakes tasks, such as sending a simple message or placing a straightforward order, but stability, app permissions, and trust remain the biggest barriers. Reviewers note that a wrongly sent message or a mistaken purchase carries real cost, so most users keep it on a short leash and supervise anything involving money or communication.
AutoGLM fits users inside the Chinese Android ecosystem who want to automate routine phone chores, and developers anywhere who want an open-source foundation for building phone agents. It is far less relevant for users whose daily apps are Google, WhatsApp, and Instagram, since its app coverage and accessibility-service integration are built around Chinese services.
Summary of public user & expert reviews, compiled by RECATOOLS.
About this listing
This entry was compiled from publicly available data including AutoGLM (Zhipu)'s official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with AutoGLM (Zhipu) unless explicitly stated.
Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.
For the latest details, please refer to AutoGLM (Zhipu) directly →
Spotted something out of date? Suggest an update →
Alternatives to AutoGLM (Zhipu)
More in Agents & Automation
