New: Agent Bridge is live

GLM Agents.

GLM Agents pair the latest models and priority access with up to 40 parallel subagents, a cloud workspace that keeps running while your PC is off, and a VS Code-style bridge to your own files.

Get started, $15/mo

See how it works

Welcome back

How can I help you today, Alex?

Ask anything, or hand off a 24-hour task

Research

Support Ops

Writing

Actions

Spin up 12 subagents to refactor the payments module

Keep a 24-hour task running in the cloud while I sleep

Draft the launch plan with the latest GLM model tonight

Fix my flaky tests through the VS Code-style bridge

Intro

GLM-workspace pairs the latest GLM models with priority access and large token limits, so every conversation starts fast and goes deep.

Up to 40 subagents run in parallel, the cloud workspace keeps working with your PC off, and a single prompt can run for up to 24 hours.

A VS Code-style workbench and an approval-gated bridge let the agent work directly on your machine, with keep/revert review on every write.

GLM-workspace pairs the latest GLM models with priority access and large token limits, so every conversation starts fast and goes deep.

Up to 40 subagents run in parallel, the cloud workspace keeps working with your PC off, and a single prompt can run for up to 24 hours.

A VS Code-style workbench and an approval-gated bridge let the agent work directly on your machine, with keep/revert review on every write.

Core Features

One unified platform to runyour entire organization

GLM-workspace turns conversation into execution: chat, delegate, and ship from one native app.

Ask

Verify

Execute

Measure

Ask: the GLM-workspace chat view with conversation sidebar, GLM model picker with priority access, and an answer with source citations.

Verify: the review working set with a diff hunk and keep-or-revert actions, and a cited web-search answer verified against live sources.

Execute: the workspace with file tree, agent steps running tests, and an approval prompt for a dangerous command.

Measure: monitoring from the companion mobile app with overnight session progress, subagent count, and a remote prompt.

Latest GLM models, priority access, live web search, and large token limits from the first prompt.

Ask

Verify

Execute

Measure

Latest GLM models, priority access, live web search, and large token limits from the first prompt.

Ask

Why GLM-workspace

GLM Agents foranswers, actions, and momentum

The full power of GLM: chat, agents, and workspace in one native Windows app.

One workspace

Chat, files, and up to 40 parallel subagents stay in one native Windows workspace.

Grounded answers

Live web search with cited sources on every answer, plus diffs on every agent write.

Real execution

From prompt to pull request: commands run on your Windows files through the bridge, edits and drafts in one place.

Approval-gated

Every command and file write on your Windows machine waits for your confirm; dangerous ones never auto-run.

Always in reach

Monitor sessions and send prompts from the companion mobile app while your desktop works.

One workspace

Chat, files, and up to 40 parallel subagents stay in one native Windows workspace.

Grounded answers

Live web search with cited sources on every answer, plus diffs on every agent write.

Real execution

From prompt to pull request: commands run on your Windows files through the bridge, edits and drafts in one place.

Approval-gated

Every command and file write on your Windows machine waits for your confirm; dangerous ones never auto-run.

Always in reach

Monitor sessions and send prompts from the companion mobile app while your desktop works.

One workspace

Chat, files, and up to 40 parallel subagents stay in one native Windows workspace.

Grounded answers

Live web search with cited sources on every answer, plus diffs on every agent write.

Real execution

From prompt to pull request: commands run on your Windows files through the bridge, edits and drafts in one place.

Approval-gated

Every command and file write on your Windows machine waits for your confirm; dangerous ones never auto-run.

Always in reach

Monitor sessions and send prompts from the companion mobile app while your desktop works.

What you get

Ask a question and get real work back

GLM-workspace takes you from question to shipped work, with your agents' whole workspace one tab away.

Cloud workspace

A workspace that keeps running while your PC is off.

The cloud sandbox comes with 2 CPU cores and 4 GB of RAM, working independently of your machine.

A single prompt can run for up to 24 hours.

Cloud workspace24h session

Files

src

payments.ts

bridge.ts

tests

api.test.ts

README.md

Refactor payments module12 subagents

test-runner · 214 passed

migrations · committed

api-client · generating

Task progress78%

Your PC is off. The workspace keeps working.

Cloud workspace24h session

Files

src

payments.ts

bridge.ts

tests

api.test.ts

README.md

Refactor payments module12 subagents

test-runner · 214 passed

migrations · committed

api-client · generating

Task progress78%

Your PC is off. The workspace keeps working.

DESKTOP-7F3K · pairedask mode

api/payments.tsKeepRevert

export async function charge(cart) {

− const total = cart.sum();

+ const total = sum(cart.items);

const fee = calcFee(total);

+ await audit('charge', total);

return gateway.pay(total);

Review working set · 3 files · 5 hunks

Approve command on your files?npm run test:payments

DenyAllow

No answer in 2 minutes counts as a rejection.

DESKTOP-7F3K · pairedask mode

api/payments.tsKeepRevert

export async function charge(cart) {

− const total = cart.sum();

+ const total = sum(cart.items);

const fee = calcFee(total);

+ await audit('charge', total);

return gateway.pay(total);

Review working set · 3 files · 5 hunks

Approve command on your files?npm run test:payments

DenyAllow

No answer in 2 minutes counts as a rejection.

Local bridge

Agents that can work on your own files.

The bridge connects the agent to your Windows machine, gated by approvals and a keep-or-revert review set.

Every write lands in a review queue first.

Companion app

Monitor and prompt from your phone.

The companion mobile app shows session progress and subagent counts, and sends new prompts while you are away.

Long-horizon work stays within reach.

23:47

Overnight session

Payments refactor12 subagents

78% · est. done 06:40

Cloud CPU

34%

Memory

61%

Sent 23:12 · from your phone

Keep the migrations rolling, I'll review at 9am

3 files added to review set

Migrations running

Send a prompt…

23:47

Overnight session

Payments refactor12 subagents

78% · est. done 06:40

Cloud CPU

34%

Memory

61%

Sent 23:12 · from your phone

Keep the migrations rolling, I'll review at 9am

3 files added to review set

Migrations running

Send a prompt…

Integrations

Connect the systems behind your workflow

GLM-workspace works with your stack: GitHub for durable work, your local files through the bridge, and the cloud tools your agents run.

GitHub connected

Agents submit finished work to your repo.

GitHub connected

Agents submit finished work to your repo.

GitHub connected

Agents submit finished work to your repo.

Local bridge

The agent works on your own files, gated.

Local bridge

The agent works on your own files, gated.

Local bridge

The agent works on your own files, gated.

Approval gate

Confirm every command and file write.

Approval gate

Confirm every command and file write.

Approval gate

Confirm every command and file write.

Review and revert

Keep or revert every change, per hunk.

Review and revert

Keep or revert every change, per hunk.

Review and revert

Keep or revert every change, per hunk.

Changelog

New features, fixes, and improvements

See all updates →

Agent Bridge 2.0 and the review working set

The agent can now work directly on your Windows machine through the approval-gated bridge. Every file write lands in a keep/revert review set: accept or revert per hunk, and your manual saves sync into the same queue.

Dangerous commands never auto-run, auto-allow lists stay local to your device, and every remote call is rate-limited and fully logged, and the CLI reports exactly what ran and why.

Mar 23, 2026

Agent Bridge 2.0 and the review working set

Dangerous commands never auto-run, auto-allow lists stay local to your device, and every remote call is rate-limited and fully logged, and the CLI reports exactly what ran and why.

Mar 23, 2026

40 parallel subagents

Complex tasks now fan out across up to 40 subagents running in parallel, all following the same model picker as your chat. Delegation, tool calls, and todo lists render live in the conversation so you can follow along.

On heavy workload, subagents may not be completely available; the main agent keeps going and picks them up again as capacity frees up. Better and more accurate results on complex work is the goal; raw speed is not guaranteed.

Subagents share the workspace's 50 GB upload budget, so large research sets and datasets move over once and stay available to every agent in the session.

Mar 4, 2026

40 parallel subagents

Subagents share the workspace's 50 GB upload budget, so large research sets and datasets move over once and stay available to every agent in the session.

Mar 4, 2026

24-hour long-horizon runtime

A single prompt can now keep working for up to 24 hours. Your cloud workspace (2 CPU / 4 GB) keeps running with your PC off, and the companion mobile app lets you monitor progress and send follow-up prompts from anywhere.

Sessions, terminal output, and file changes stream back as they happen, so a long run never turns into a black box. If something needs your call, the agent asks and waits.

Feb 10, 2026

24-hour long-horizon runtime

Sessions, terminal output, and file changes stream back as they happen, so a long run never turns into a black box. If something needs your call, the agent asks and waits.

Feb 10, 2026

GitHub-connected persistence

The cloud sandbox is powerful but not permanent: environments reset between sessions. Connect GitHub access and agents submit completed work to your repository as commits, branches, and pull requests, even while you're offline.

Work saved on the chat always persists, but the files live in the environment. GitHub is the bridge between the two: the durable home for everything your agents finish.

Pair it with the local bridge and you get the full loop: cloud work lands in your repo, local work lands on your disk, and you review both in one place.

Jan 20, 2026

GitHub-connected persistence

Work saved on the chat always persists, but the files live in the environment. GitHub is the bridge between the two: the durable home for everything your agents finish.

Pair it with the local bridge and you get the full loop: cloud work lands in your repo, local work lands on your disk, and you review both in one place.

Jan 20, 2026

Numbers

Every prompt, file, and subagent adds up to shipped work

40

Subagents running at the same time inside one workspace, on the tasks you delegate.

Parallel subagents

24h

The longest a single prompt can keep working in the cloud workspace, even with your PC off.

Max prompt runtime

Per-file upload limit, room for datasets, video, and archives

Dedicated cores with 4 GB RAM in every cloud workspace

Of completed remote work delivered to your GitHub repository

One subscription for the desktop app and companion mobile app

Pricing

One plan. Everything included.

No tiers, no upsells. $15 a month gets you the whole workspace, on desktop and mobile.

LICENSE

Monthly

/month

The complete workspace: Windows app plus the companion mobile app.

Get GLM-workspace

Latest GLM models

Priority access

Up to 40 parallel subagents

Cloud workspace (2 CPU / 4 GB) that keeps working with your PC off

50 GB data upload limit

VS Code-style workbench with approval-gated agent bridge

Companion mobile app for monitoring and remote prompting

Up to 24 hours of runtime on a single prompt

Large token limits

LICENSE

Monthly

/month

The complete workspace: Windows app plus the companion mobile app.

Get GLM-workspace

Latest GLM models

Priority access

Up to 40 parallel subagents

Cloud workspace (2 CPU / 4 GB) that keeps working with your PC off

50 GB data upload limit

VS Code-style workbench with approval-gated agent bridge

Companion mobile app for monitoring and remote prompting

Up to 24 hours of runtime on a single prompt

Large token limits

LICENSE

Monthly

/month

The complete workspace: Windows app plus the companion mobile app.

Get GLM-workspace

Latest GLM models

Priority access

Up to 40 parallel subagents

Cloud workspace (2 CPU / 4 GB) that keeps working with your PC off

50 GB data upload limit

VS Code-style workbench with approval-gated agent bridge

Companion mobile app for monitoring and remote prompting

Up to 24 hours of runtime on a single prompt

Large token limits

Fair-use licensing. Your subscription covers the Windows app and the companion mobile app: one Windows device and one mobile device, locked to prevent abuse and bulk sharing. The service is provided for as long as Z.ai's service remains active. If a cutoff ever interrupts it, you can request a refund.

FAQ

Answers to the questions that come up most.

Learn how GLM-workspace works, what it connects to, how the bridge is gated, and what to expect day to day.

General

Subagents & Workspaces

Licensing & Billing

Still have questions?

Can’t find what you’re looking for? The features section covers the workspace, subagents, and the bridge in more depth, and the changelog tracks every update.

Explore the features →

What is GLM-workspace?

A native Windows app that brings the GLM model to your desktop: the chat.z.ai experience with a VS Code-style workspace, priority access, and up-to-date GLM models.

Which models can I use?

All the latest GLM models, through the same model picker you already use in chat. Subagents use that picker too, so every task runs on the model you choose.

Does it keep working when my PC is off?

Yes. Your cloud workspace (2 CPU / 4 GB) keeps running on its own, so a single prompt can keep working for up to 24 hours without your machine being open.

How do I get completed work back?

Connect GitHub access to the agent. When a task finishes while you are away, the agent can submit the changes to your GitHub repository directly.

Is there a mobile app?

Yes. The subscription includes a companion mobile app for monitoring your agents and sending prompts remotely, so long-horizon tasks stay within reach even when you are away from your PC.

General

Subagents & Workspaces

Licensing & Billing

Still have questions?

Can’t find what you’re looking for? The features section covers the workspace, subagents, and the bridge in more depth, and the changelog tracks every update.

Explore the features →

What is GLM-workspace?

A native Windows app that brings the GLM model to your desktop: the chat.z.ai experience with a VS Code-style workspace, priority access, and up-to-date GLM models.

Which models can I use?

All the latest GLM models, through the same model picker you already use in chat. Subagents use that picker too, so every task runs on the model you choose.

Does it keep working when my PC is off?

Yes. Your cloud workspace (2 CPU / 4 GB) keeps running on its own, so a single prompt can keep working for up to 24 hours without your machine being open.

How do I get completed work back?

Connect GitHub access to the agent. When a task finishes while you are away, the agent can submit the changes to your GitHub repository directly.

Is there a mobile app?

Yes. The subscription includes a companion mobile app for monitoring your agents and sending prompts remotely, so long-horizon tasks stay within reach even when you are away from your PC.

General

Subagents & Workspaces

Licensing & Billing

What is GLM-workspace?

A native Windows app that brings the GLM model to your desktop: the chat.z.ai experience with a VS Code-style workspace, priority access, and up-to-date GLM models.

Which models can I use?

All the latest GLM models, through the same model picker you already use in chat. Subagents use that picker too, so every task runs on the model you choose.

Does it keep working when my PC is off?

Yes. Your cloud workspace (2 CPU / 4 GB) keeps running on its own, so a single prompt can keep working for up to 24 hours without your machine being open.

How do I get completed work back?

Connect GitHub access to the agent. When a task finishes while you are away, the agent can submit the changes to your GitHub repository directly.

Is there a mobile app?

Yes. The subscription includes a companion mobile app for monitoring your agents and sending prompts remotely, so long-horizon tasks stay within reach even when you are away from your PC.

Testimonials