Pricing

One balance, spent on whole jobs.

Run time costs $0.01 per minute, billed by the second, and model usage bills at cost with no markup. Our task-cost benchmark on v349, 14 October 2026, used started-minute compute billing. A run of a repository's test suite cost $0.027 at the median, and a research task saved as notes cost $0.055. How each run is billed is in What a credit buys.

Workflows, priced per run.

A workflow is an automation your plori agent builds for you: it runs on a schedule, on a webhook, or on demand, without an agent turn. A workflow has no fee of its own until it runs; the agent turns that build or edit it bill like other turns. You pay a small flat fee when the workflow actually runs, shown here straight from the billing system. AI steps bill their model usage on top at the model's own rate, so the flat fee is not the total for workflows that call a model.

Every workflow also carries its own monthly spend cap, so a runaway schedule can't spend your credits unnoticed: when the cap or your balance runs out, fires are skipped and recorded, never billed. Plans differ in how many workflows can be active at once (1 on Free, 5 on Pro, 20 on Power). Run length is not a plan feature: every run on every plan may go up to a 5-minute runaway protection stop, and pays only for the run time it actually used.

Questions, answered.

How do the plans work?

Each plan adds a monthly amount of credits, in US dollars, and includes one account-wide disk that all your agents share. Free includes $5 of monthly usage and 1 GB; Pro is $20/month including $20 of usage and 20 GB; Power is $100/month including $120 of usage and 50 GB. Your monthly usage refreshes each cycle and doesn't roll over; you can top up extra credit (which doesn't expire) or buy more disk anytime.

What is a credit?

Your balance is one number, held in US dollars, that covers everything an agent does: thinking, run time, and the disk it keeps beyond your plan. You watch one number, never converting tokens. It is spent on whole jobs, and each job's cost already includes both the thinking and the run time, so there are no separate budgets to reconcile. Model usage is billed at cost with no markup, and active compute meters at $0.01 per minute of run time, billed by the second; the balance is just those meters rolled into one number.

Is run time billed?

Yes. On every plan, run time costs $0.01 per minute, billed by the second. Plori charges it to the same dollar balance as model usage. A plan gives you a monthly balance in dollars, not a number of minutes. Model usage draws credits unless Plori applies an automatic platform-fault refund. Idle agents sleep and stop billing for compute.

Which models can I use on each plan?

Every plan uses the same models. You pick the Plori Router and it picks one model per request from a shared pool. We choose the pool for quality per dollar and change it as models are released and repriced, so this page does not list them. Each run's usage shows the model that served it and what it cost. The router does not read your task to choose a model. Model usage bills at what the model costs us, with no markup, and is $0 with your own key. Bring your own OpenRouter or provider key to use a specific model instead of the router. That usage bills your provider directly instead of drawing credits.

How does disk work, and can I buy more?

Your account gets one disk allowance (1 GB on Free, 20 GB on Pro, 50 GB on Power), and the files of all your agents count against it. Disk within your plan is free. Need more? Buy extra by the gigabyte at $0.10 per GB per month, on any plan including Free; it raises your account's disk and is billed from your balance while you keep it. Slide the estimator above to see the monthly cost, then buy it from your dashboard.

Is buying disk a one-time or a monthly charge?

Monthly, prorated daily. You pay $0.10 per GB per month for the extra disk you keep, drawn from your balance a little each day, for as long as you keep it. Lower it or drop back to your plan's included disk anytime and the charge stops; you're billed for what you reserve, not how full it is.

What happens when the disk fills up?

Your account's disk is a hard cap: included plus whatever you've bought. When your agents' files reach it, further writes stop (like any full disk) rather than running up a surprise bill. Free up space or buy more disk and they continue. All your agents share this one disk, so any of them can use the whole thing.

What do I pay while my agent is asleep?

Usually nothing. A plori agent scales to zero when idle: compute stops, and the disk within your plan is free, so a sleeping agent that fits its included disk costs zero while keeping its files and memory intact. You only pay while asleep if you've bought extra disk above your plan.

Do unused credits roll over?

Your plan's monthly credits don't roll over; they refresh at the start of each billing cycle, so they're meant to be used within the month. Credits you buy separately as a top-up, and your one-time welcome credit, are different: they never expire and stay on your balance until you spend them.

What do built-in tools cost?

Nothing. Built-in tools, including web search, X and Reddit reads, and the SEO tools, are free to call. You pay only for the run time and model usage of the job that calls them. Each tool has a daily call limit per account, which resets at 00:00 UTC and stops a runaway run from calling it without end. The limits are listed at api.plori.ai/v1/pricing/tools.

What do workflows cost?

A workflow has no fee of its own until it runs. The agent turns that build or edit it bill like other turns. When a workflow runs, it draws a flat per-run fee from your balance with the first minute and a half of run time included, then a small per-minute charge for longer runs; the exact rates are in the Workflows table above, served straight from the billing system. If the workflow has an AI step, its model usage bills on top at the model's own rate, at cost with no markup (no charge with your own provider key), so the flat fee alone is not the total for workflows that call a model. You're only charged when your workflow actually ran: it succeeded, failed, or hit the 5-minute runaway protection stop (a system safety limit, the same on every plan; a run stopped there pays for the run time it used). Skipped or platform-lost fires are free, and each workflow has a monthly spend cap you control.

Can I use my own model API key?

Yes. Add your own provider key in dashboard settings and select a model from that provider. Its model calls then stop drawing credits: you pay the provider directly. Compute, extra disk and workflow runs still bill from your balance. If you keep the Plori Router, model calls can still bill from your balance. A provider key is the cheapest way to run heavy frontier workloads.

What happens if I run out of credits?

Your plori agent stops taking new work at zero, so there's no surprise bill. Its disk and memory are kept at no charge for a 30-day grace period. After that, the disk is archived offline and is not deleted. When you top up or your monthly credits renew, Plori restores the disk and the agent resumes where it stopped.

The sleep answer is metered, not promised: the sleep-billing benchmark publishes the ledger of a working session and the sleep window after it.

Open an agent and watch the balance.