Opening soon · Nova, Ohio

Hosting your agent can sign up for.

Tell your coding agent to deploy to PloyHQ. It makes the account, asks you to pay once, and hands back a live HTTPS URL. No dashboard required.

Read llms.txt See prices

you › ship this to ployhq

POST/v1/accounts→ api_key phq_7f3c… saved to PLOYHQ_API_KEY

POST/v1/credits {"amount_usd": 20}→ checkout link sent to you · paid

POST/v1/apps {"name": "todo-api"}→ status provisioning

GET/v1/apps/app_91c0…→ status running https://todo-api.ployhq.app

How it works

Four calls from nothing to a URL

Make an account
No form and no email confirmation. The key is shown once, so the agent saves it.
```
POST /v1/accounts
{"email": "optional"}
```
Add credit
The agent gets a checkout link and passes it to you. $20 is the minimum and it's the only step that needs a human.
```
POST /v1/credits
{"amount_usd": 20}
```

Deploy

From a git repo or a container image. Every app gets its own VM and a URL on ployhq.app.

POST /v1/apps
{"name": "todo-api", "plan": "small",
 "source": {"git": "https://github.com/you/todo"}}

Poll until running

Then open the URL. Logs, balance and runway are one GET each.

GET /v1/apps/{id}
{"status": "running", "url": "https://todo-api.ployhq.app"}

Pricing

About $3 per GB of RAM a month

Per second

Billed only while an app runs. $3 a month is $0.0000011 a second.

$20 to start

Prepaid credit. No free tier, no surprise invoice.

Stops at $0

Apps pause when credit runs out and keep their disk until you top up.

Plan	RAM	Shared vCPU	Disk	Standard (default)	Classic
nano	512 MB	1	10 GB	$1.50/mo$0.0021/hr	$1.00/mo$0.0014/hr
small	1 GB	1	20 GB	$3.00/mo$0.0041/hr	$2.00/mo$0.0027/hr
medium	2 GB	2	40 GB	$6.00/mo$0.0082/hr	$4.00/mo$0.0055/hr
large	4 GB	2	60 GB	$11.00/mo$0.0151/hr	$7.50/mo$0.0103/hr
xl	8 GB	4	100 GB	$20.00/mo$0.0274/hr	$13.50/mo$0.0185/hr

CPU tiers

Pick by instruction set, not by brand

Standard Default

Intel Xeon E5-2697 v4 · AVX2

Runs anything built for modern x86-64, including prebuilt Bun, Node native modules and AI/ML Python wheels.

Leave tier out and this is what you get.

Classic

Intel Xeon E5-2680 v2 · AVX, no AVX2

About a third cheaper. Good for plain web apps, Go and Python without compiled AVX2 wheels.

Software that needs AVX2 exits with illegal instruction here. Send "tier": "classic" only when you know it doesn't.

Errors

Every error says what to do next

Agents recover on their own when the error names the fix. Every response that isn't a success carries a fix field written for the agent to act on.

HTTP/1.1 402 Payment Required
{"error": {
  "code": "insufficient_credit",
  "message": "Balance $0.00 is under the $0.10 needed for 24 hours of standard plan 'small'.",
  "fix": "POST /v1/credits with {\"amount_usd\": 20} and give the checkout link to the user."
}}

For agents

Everything is machine-readable

/llms.txtThe whole product in one page of plain text: the four calls, prices and gotchas.

/openapi.jsonOpenAPI 3.1 spec for every endpoint, including error shapes.

/pricing.jsonEvery plan and tier with monthly, hourly and per-second prices.

MCP server

Nine tools that mirror the API. Add it to Claude Code:

claude mcp add ployhq -e PLOYHQ_API_KEY=phq_… -- python3 -m ployhq.mcp_server