We just launched on Product Hunt

Virtual Machines
for Agents

Just ask Claude, ChatGPT, or Cursor. Get a VM.

Try it on Claude Open the Console →
AMD EPYC
Dedicated vCPU
NVMe
Unthrottled storage
Unlimited
Egress, no markup
Per second
Billing
Speaks MCP →
Claude Code Claude Desktop Cursor Antigravity Windsurf Zed

Skip the docs.
Skip the dashboard.

First-timers

Publish what your AI built

"host my python script on a VM"
Builders

Let your agents self-deploy

"deploy each generated app to its own box"
Try it on Claude Open the Console →

Three steps. One terminal.

01
Copy the install line
02
Paste it into your terminal
03
Chat to deploy
$ claude mcp add huddle01 -- npx -y @huddle01/mcp
Try it on Claude Open the Console →

Faster. 1/3 the price.
No Manual Configuration.

30+ min sustained · Ubuntu 24.04 LTS (Kernel 6.8) · ffmpeg, Go, make, fio · Open-source · AP-South-1

Web Server — Go, 50 concurrent image requests
Huddle01 anton-2
84.26 RPS
AWS c7i.large
46.18 RPS
3× the cost
AWS t3.medium
28.44 RPS
CRASHED
Video Transcoding — 4K → 1080p (lower is better)
Huddle01 anton-2
13 mins
AWS c7i.large
19 mins
AWS t3.medium
32 mins
Throttled
CI/CD Build — Redis compilation from source (lower is better)
Huddle01 anton-2
67 sec
AWS c7i.large
84 sec
AWS t3.medium
128 sec
Disk I/O — PostgreSQL-style random read/write (IOPS — higher is better)
Huddle01 anton-2
13,380
AWS c7i.large
2,286
AWS t3.medium
2,266
5× the IOPS. 7× lower latency. At 1/3 the price.
Try it on Claude Open the Console →

Just type it.
Watch it happen.

01 first-timer
"set up a VM for my python script"
02 builder
"deploy this Next.js repo to a fresh box"
03 either
"how much have my VMs cost this week?"
Try it on Claude Open the Console →

No brand tax.
No egress markup.

~70%
more affordable than hyperscalers
ClassComputeMemoryStorageFrom
Standard4 vCPU · AMD EPYC16 GB160 GB NVMe$0.018/hr
Performance8 vCPU · AMD EPYC32 GB320 GB NVMe$0.072/hr
Compute+16 vCPU · AMD EPYC64 GB640 GB NVMe$0.144/hr
Memory+8 vCPU · AMD EPYC128 GB640 GB NVMe$0.260/hr
Per-second billing. Unlimited egress on every tier. See full price list →
Try it on Claude Open the Console →

The actual
questions.

Two crowds. Non-developers who want to publish what they're building with an AI assistant, without learning ssh, security groups, and dashboards. And builders who want their multi-agent systems to self-deploy the apps they generate.
We're MCP-native. The first VM provider that lets your agent provision infra by chatting. There's no brand tax: you pay for compute, full stop. No egress markup. In practice that lands ~70% below the hyperscalers.
Dedicated AMD EPYC vCPUs, NVMe-backed storage with no throttling, and unlimited egress across global regions. You get an SSH key, a public IP, and a URL on huddle01.app.
Anything that speaks Model Context Protocol: Claude Code, Claude Desktop, Cursor, Antigravity, Windsurf, Zed, Continue. If your tool supports MCP, it supports Huddle01 VMs.
Per second. The moment your VM stops, billing stops. No idle charges, no minimum commitment, no egress surprises.
Yes. console.huddle01.com gives you the full UI, and there's a CLI for scripts and CI. The MCP server is the third path. Pick whichever fits the moment.

Stop clicking.
Start chatting.

Install the MCP server Upvote on Product Hunt ↗ Join the Discord →