Transparent pricing. Every task shows every dollar.

Rate cards are theater. They tell you the list price of a token; they don't tell you what your task cost. So we don't lead with one.Every task on Rebyte finishes with a line-item receipt — compute hours, model calls, skill calls, totals. You don't have to trust a rate; you see the spend, per task, in credits and in dollars.
"We bill what we cost, and we show you the bill on every task."

Three ways to pay for models

Model spend is almost always the largest line on any agent bill. You pick the relationship that fits how your team already buys AI.

Bring your own API key

Point Rebyte at your Anthropic, OpenAI, or Google key. Calls hit your provider contract directly. You see their invoice; Rebyte takes no margin on tokens.

Bring your own subscription

Use your existing Claude Max or Gemini subscription to drive agents. Nothing extra from us for model spend.

Rebyte credits

Prefer one bill? Use our hosted models at at-cost token rates. Every task shows exactly how many credits each model call consumed.

Every task comes with a receipt

Transparency is not a rate card pinned to a marketing page — it is the itemized bill you get after the work is done. On Rebyte, every task ends with:

Compute

Hours the agent computer ran. $0.10/hr while active, $0 when idle.

Model calls

Each call, which model, input and output tokens, credits (or $0 on BYOK / BYOS).

Proprietary skills

Image generation, video generation, and other hosted data services — per request, itemized.

Total

Credits spent and dollars equivalent, for the task as a whole.
No bundled "AI tokens" hiding what you actually used. No opaque overage charges. Open any task, scroll to the bottom, read the bill.

Plans

Compute is not a billing gate. Every plan gets unlimited agent computers; the upgrade buys governance, storage, and connectors.

Personal

$0
Daily free usage · pay-as-you-go credits

Pro

$20 /mo
Bring your own key · private skills · full history

Team

$20 /seat/mo
Governance · shared workspaces · team collaboration · Agent Context

Enterprise

—
SSO · SCIM · dedicated VPC · on-prem models · custom SLAs
Start freeBook a call

Questions

Why don't you publish a per-token rate card?

Because the rate is meaningless without the usage. Instead, we pass through at-cost rates for hosted models and show the exact credit spend on every task. If you want rates in advance, bring your own key — the provider publishes them.

What counts as a "proprietary skill"?

Anything that costs us money beyond tokens — image generation, video generation, other hosted data services. These are always metered per request and listed on the task receipt.

Which subscriptions can I bring?

Claude Max and Gemini, today. Drive your agents with the plan you already pay for; we don't charge for the tokens.

Is there really a free tier?

Yes. Daily free usage on every agent, no credit card. When you outgrow it, pay-as-you-go credits, or bring your own key or subscription.

Do you count tasks or machines?

Neither. Agent computers and concurrent tasks are both unlimited. We bill the human seat (on Team), compute hours, and model or skill usage — each shown per task.

Can I change my plan later?

Yes. Upgrade when you need governance or connectors; add or remove seats whenever. Billing is prorated.