Limits & pricing

What you pay

There are three separate things, and it helps to keep them apart:

Cost	Goes to	Amount
Endpoint price	The endpoint owner	Whatever they quote
Network fee	Algorand	~0.001 ALGO per transaction
Platform fee	Ripar	A percentage of settled endpoint revenue

Ripar takes its cut from settled revenue on the marketplace. Deploying, running workflows and calling endpoints do not carry a per-seat or per-call platform charge.

No seats

Pricing does not scale with how many people are on your team. It scales with how much your endpoints earn.

Platform limits

Limit	Value	Notes
Endpoint timeout	300s max	Default 30s
Request body	10 MB	Larger payloads should use a URL
Response body	10 MB	Same
Workflow steps	100 per run	Split larger jobs
Workflow duration	15 min per run	Cron workflows can chain
Management API	600 req/min	Per token
Idempotency window	24h	Keys are remembered this long
Job budget	0.10 USDC min	No maximum
Bid window	1 min – 24h	Set with `--closes`

Cold starts

Node templates typically boot in under 300ms. Endpoints with steady traffic stay warm. An endpoint that must never cold-start can hold a minimum instance:

ripar scale summarize --min-instances 1

That instance runs continuously, so it costs whether or not anyone calls.

Rate limiting your own endpoint

Ripar does not throttle paid calls — payment is the throttle. If you want a hard ceiling regardless:

rateLimit: { perMinute: 600, per: "payer" },

Exceeding it returns 429 and does not charge the caller.

Quotas

Hobby accounts have a monthly cap on settled call volume before a plan upgrade is required. The dashboard shows current usage; ripar limits show prints it too.