Claude Desktop System Requirements: Windows, macOS, Linux (2026)

Heavy AI use has changed shape in the last six months. Claude Desktop is no longer the chat-with-an-API app it was a year ago. Claude Cowork ships an agentic layer alongside the chat tab. The Model Context Protocol install dance is now a one-click marketplace . The default model is Opus 4.8 on the heavy tier and Sonnet 4.6 on the daily tier. The hardware question has shifted along with the product.

I run Claude Desktop daily alongside Claude Code (the CLI) and a stack of MCP servers I publish myself under @houtini on npm. This guide is what I'd tell anyone speccing a machine for the same workflow in June 2026: what's official, what's optimistic, where the install hits friction, and what you need to make it work.

What Claude Desktop contains in 2026

Claude Desktop is the Electron-based app you download from claude.com/download . Inside it, three modes share the same shell:

1. Chat - the familiar conversational interface. What most people opened the app to use a year ago. Still here, still the default tab.
2. Cowork - the agentic mode. You point it at a folder, give it a goal, and it works through the task using local files and connected tools. Released on Mac in research preview January 12, 2026 , generally available on macOS and Windows on April 9, 2026 .
3. Claude Code - the CLI dev tool now bundled into the desktop app rather than a separate install. See the dedicated Claude Code system requirements article for the developer side.

Three modes, one binary, one set of system requirements to think about. The base requirement is set by Chat; Cowork raises the bar; Claude Code is roughly Chat-tier for the CLI itself with sub-agents adding bursty load.

This article is about Claude Desktop. If you came looking for Claude Code (the terminal CLI), the dedicated guide is linked above. The browser-based Claude.ai is over at claude.ai . Both work great for what they are - this one is for the desktop app.

Official requirements

Anthropic publishes the canonical minimums on the download page and across the release notes . The short version:

The chat-only minimum holds in practice. The moment Cowork starts running, the numbers stop being academic.

Cowork: the new agentic layer

Cowork is the single biggest shift to Claude Desktop in the last year. It is worth understanding before you decide whether your machine can run it.

A scope note before I dig in: my own agentic work lives in Claude Code on the CLI side, not in Cowork. I've spun up Cowork sessions to confirm the install path works and to watch the VM behaviour, but I haven't put it through heavy sustained client work the way I have with Claude Code. The architecture below comes from Anthropic's published documentation and release notes, cross-checked against the practitioner reports I trust; my own contribution is interpreting what those numbers mean for a machine you're about to spec.

What Cowork is. A mode that takes a goal and works through it. You point it at a folder, describe the outcome, and Cowork plans the steps, edits files, runs tools, and returns a finished deliverable. Anthropic's framing is precise: this is not a chat assistant. It is an agentic system that executes multi-step knowledge work on your behalf.

How it runs. Cowork executes in an isolated virtual machine on your computer. The release-note statement is verbatim: "It runs locally on your computer in an isolated VM, enabling direct access to local files and MCP integrations." This isolation is the safety feature. It also explains the resource footprint: Cowork is heavier than Chat because there is a VM to spin up before any work begins.

What plans get it. Pro, Max, Team, and Enterprise. Free is excluded. The expansion happened in stages:

• 12 January 2026 - research preview, macOS only, Max plan only
• 16 January 2026 - expanded to Pro plan, still macOS only
• 9 April 2026 - general availability on macOS and Windows for all paid plans
• 24 February 2026 - plugin marketplace for Team and Enterprise
• 25 February 2026 - scheduled recurring tasks and a new Customize section grouping skills, plugins, and connectors
• 17 March 2026 - Dispatch from your phone (Pro/Max research preview)
• 23 March 2026 - computer use in Cowork and Claude Code

What Cowork demands of your machine. The VM consumes resources only while it is running. An idle Claude Desktop with the Cowork tab unopened is no more demanding than the Chat-only version. The moment you start a Cowork session, the VM boots, the work-trees get created (one per parallel sub-agent), and the resource usage steps up.

A practical floor for serious Cowork use:

• 16GB RAM minimum if you run other dev tools alongside (browser, editor, Slack)
• 5GB free disk for the Cowork VM image plus your project work-trees
• Hardware virtualisation enabled in BIOS (VT-x on Intel, AMD-V on AMD) - without it the Windows VM stack refuses to boot
• Wired-quality network connection when running long Cowork tasks - model inference happens in Anthropic's cloud, so every step is a round trip

What Cowork is NOT. Worth being explicit because the marketing pivots can mislead.

• Not a local LLM. All inference still happens in Anthropic's cloud. The VM runs the agent loop and the file operations, not the model itself.
• Not fully autonomous. Cowork proposes a plan and waits for your approval before taking action. The phrase Anthropic uses is "you're delegating the work, not the judgment." Approval gates are first-class.
• Not for free-tier users. The download page works for everyone; the Cowork tab is gated to paid plans.

Connectors: how you add MCP servers now

The hardest part of running Claude Desktop a year ago was wiring up MCP servers. You installed Node.js, you edited claude_desktop_config.json by hand, you wrestled with dependencies, you restarted the app, you hoped. That whole flow has been replaced.

I publish a suite of MCP servers under @houtini on npm, so I've watched this whole install-friction story from the developer side. When the only install path was a JSON config dance, adoption capped out at the developers willing to wrestle Node and read changelog notes. The marketplace shift changes the addressable audience for MCPs entirely - non-developers can now install one in two clicks, and that's the bit nobody flagged at the time as the actual story of 2026.

There are now three install paths, in increasing complexity:

1. The Connectors directory (the default path). Browse the catalogue, click Connect on a remote service like Notion or Canva or Stripe, complete the OAuth flow, done. No terminal, no config files, no dependency conflicts. The directory launched on 14 July 2025 and has grown steadily since.

2. Desktop Extensions (.mcpb files). For local MCP servers - file system access, database queries, custom local tools. Download a .mcpb file, double-click to open with Claude Desktop, click Install. Built-in Node.js runtime ships with the app so the historic dependency dance is gone. Secrets land in your OS keychain. Updates happen automatically.

Anthropic's own framing of the shift, verbatim :

Before: Install Node.js first → npm install -g @example/mcp-server → Edit ~/.claude/claude_desktop_config.json manually → Restart Claude Desktop → Hope it works After: Download a .mcpb file → Double-click to open with Claude Desktop → Click "Install"

3. Manual claude_desktop_config.json editing (the escape hatch). Still works. Still required for arbitrary GitHub-published MCP servers that haven't been packaged as .mcpb files. Useful for development, internal tools, or anything experimental. The marketplace is the front door; manual configuration is the back door. Both stay open.

System-requirements implications:

• Remote connectors (Notion, Stripe, Google Drive, anything OAuth) - negligible local resource cost. The connector calls run in Anthropic's cloud, not on your machine. Network policy matters; processing power doesn't.
• Local desktop extensions (file system, database, local tools) - small. The bundled runtime handles dependencies. Some extensions may spawn helper processes; budget RAM for those if you install many.
• Custom self-hosted MCPs via JSON config - depends on what the MCP does. If it spawns a Python process per call, it consumes whatever that process needs.

The Connectors marketplace makes adding tools a click. The system-requirements implication is small for most users; the bigger shift is that the JSON-editing skill is no longer required for participation. If you ever skipped MCPs because the install was too painful, the install is no longer painful.

What you need

When I'm specifying a machine for someone who plans to use Cowork seriously, the conversation starts somewhere different from Anthropic's published minimums. Those minimums are conservative - they describe the bottom of what works for Chat-only use - and they are not what most people running Claude Desktop in anger end up using.

A realistic picture for the three usage tiers:

Chat-only (Free / Pro / Max / Team / Enterprise - same hardware story).

• 8GB RAM is enough
• macOS 13+ or Windows 10 1809+ is enough
• Any modern CPU
• No Cowork, no Connectors-as-OS-control, no agent loops

Cowork-active (Pro / Max / Team / Enterprise).

• 16GB RAM is the floor for comfortable use alongside other dev tools
• 32GB if you run Cowork against large monorepos with parallel sub-agents
• Hardware virtualisation enabled in BIOS
• 5GB free disk minimum for the Cowork VM image
• Apple Silicon Mac (M1 or newer) gives the smoothest macOS experience; Intel Macs may struggle with the VM workload depending on age

Power user (Cowork + Claude Code + Connectors + Skills).

• 32GB RAM is the new normal
• SSD with at least 50GB free if you run Cowork against your full development workspace
• Recent CPU (8th-gen Intel or M1+ Apple Silicon equivalent)
• The bottleneck is rarely the CLI itself; it is the dev tools Claude Code runs in parallel during agentic workflows

GPU: still not needed for any of this. All model inference happens in Anthropic's cloud. The "do I need a 3090" question that comes up sometimes is usually mixing Claude Desktop up with running a local LLM - that is a different category of product, and we cover the local-LLM GPU question in Best GPUs for Running Local LLMs .

Linux: officially unsupported

There is no official Linux build of Claude Desktop. The download page offers macOS and Windows. The Cowork release notes name macOS and Windows. The Connectors marketplace works "on Claude, Claude Desktop, and Claude for iOS and Android" - Linux is absent from every Anthropic page.

If you are on Linux, your options are:

1. Use Claude Code instead. The terminal CLI runs natively on Linux. No Desktop app required for developer workflows.
2. Run Claude in your browser at claude.ai. No file-system or Connectors-marketplace integration, but Chat works fine.
3. Community ports. Open-source projects exist that package the Desktop app for various Linux distributions. They are unofficial, vary in completeness, and are not supported by Anthropic. If you go this route, treat it as a developer experiment rather than a production install.

For most Linux users, Claude Code plus the web interface covers the working envelope. Cowork specifically is the feature you cannot replicate on Linux today.

Common mistakes

Four failure modes come up over and over with Claude Desktop installs, in rough order of how often they trip people up:

1. Trying to run Cowork on the Free plan. The Claude Desktop app downloads and installs on Free, and the Chat tab works as you would expect. Cowork is gated to paid plans though, and what trips people up here is seeing the app working in Chat mode and assuming Cowork is just hidden in a menu somewhere. It is not - the gate is plan-based, not UI-based.

2. MSIX install blocked by corporate group policy on Windows. Claude Desktop ships as an MSIX package on Windows. Some corporate machines block MSIX installation by policy. The error is vague - the installer just refuses to run. The fix is to whitelist the Anthropic publisher certificate with your IT team, or use the standalone .exe installer Anthropic ships for environments where MSIX is blocked. This catches more people than any hardware problem does.

3. Hardware virtualisation disabled in BIOS on Windows. Cowork's isolated VM needs hardware virtualisation. On Windows, that means VT-x (Intel) or AMD-V (AMD) enabled in BIOS. If your IT team disabled it, or you have a custom BIOS configuration, Cowork will fail to start with an opaque error. Reboot into BIOS, enable virtualisation, save, restart, retry.

4. Closing the desktop app mid-task. Cowork tasks run while the app is open. If you quit Claude Desktop while a Cowork task is mid-run, the task pauses (or fails, depending on what it was doing). Anthropic's own guidance is direct: the desktop app needs to stay open while Cowork is working. If you set a scheduled task and quit the app, the schedule does not fire.

The fifth one worth flagging is outdated MCP installs. If you set up an MCP server via the JSON config six months ago and the same server is now in the Connectors directory or available as a .mcpb extension, your old install probably still works but is missing the marketplace's automatic updates, secret management, and uninstall hygiene. Migrate when convenient.

Where to go from here

Download from claude.com/download . On first run, sign in, choose a plan, and the app drops you in the Chat tab. From there:

• For chat-only use, you are done. Your hardware floor is the lowest of all the tiers.
• For Cowork, switch to the Cowork tab, grant access to a working folder, and start with a small task to confirm the VM boots cleanly.
• For MCP servers, open the Connectors directory at claude.ai/directory , pick a remote service to start with (Google Drive or Notion are the gentlest on-ramps), authenticate, done.
• For Linux, use Claude Code instead - or run Chat in a browser.

Claude Desktop in 2026 is a different product from the chat app it was a year ago. The hardware story is the same on the bottom and meaningfully heavier on the top. The Cowork tab is the part that changes the system requirements calculus; the Connectors marketplace is the part that makes adding capability painless. Get one machine matched to the heavier tier you use, and the rest is configuration.