46. The pricing model

Three things you pay for, one thing you don’t, and why running inference on your own GPU costs nothing per call.

This chapter is the mental model for how Daalu is priced. The numbers here are indicative — the authoritative prices live at https://daalu.io/pricing and on your invoice — but the shape of the model below changes infrequently, and it’s the shape that helps you reason about your bill.

Why it matters. Most platforms in this space meter one thing and surprise you with another. Daalu’s model is deliberately legible: you can predict your bill from three inputs, and the one input that scales fastest — AI inference — is the one you can drive to near-zero by pointing it at hardware you already own.

What you pay for

Daalu’s pricing has three components:

1. Seats — per active user, per month.
2. Connected sources — per integration, per month, after a generous included quota.
3. AI inference — metered per call, against the tier that actually served it. Pass-through on commercial models, no markup beyond a documented handling fee.

And one component you don’t pay Daalu for: inference served by your own GPU is free at the Daalu layer. You bought the hardware; we don’t charge you per call to use it. That tier is the heart of the AI Factory, and it’s why many teams see their inference line trend down as usage grows.

Plans

Three self-serve plans, plus Enterprise:

* Unlimited within reason — we’ll talk if you’re onboarding tens of thousands of users.

• Starter is for trying Daalu and very small teams.
• Team is where most customers start.
• Business is for teams running real production load.
• Enterprise adds SSO/SAML, a private region, a security-review pack, custom contract terms, and an SLA up to four nines. Talk to sales.

The exact dollar prices and the precise feature matrix live on the live pricing page.

Seat counting

A seat is a unique active user in your tenant during the billing month.

• Invited but never logged in → doesn’t count.
• Logged in but idle for the past 30 days → doesn’t count that month.
• Removed mid-month → pro-rated.

The effect: teams with seasonal usage don’t pay for dormant users, and teams where everyone is heavy don’t get artificially cheap bills. You pay for the people actually using the product.

Connected sources

A connected source is one configured integration. One AWS account = 1. One on-prem cluster = 1. One Slack workspace = 1.

The included quota covers most teams. Beyond it, each additional source is metered at a small monthly fee.

What never counts against the quota:

• Webhooks, inbound or outbound — unlimited.
• Email notifications — included.

Inference metering

This is the component that scales with how hard your team leans on the Assistant and agents — and it’s the one you have the most control over.

Every LLM call Daalu makes is billed against the tier that actually served it, not the tier you asked for. You request “classifier” or “quality”; the router decides where it lands, and you pay for the outcome:

Pass-through means Daalu doesn’t mark up the provider’s price. Your invoice shows the provider’s per-token rate × tokens consumed, plus the documented handling fee — and nothing else.

Tip. Watch your live inference mix on Usage & Pricing → Inference. The higher the share served by your own GPU and the Daalu-hosted tier, the lower the commercial line. Chapter 48 is the weekly habit that keeps it that way.

Inference is also subject to your plan’s included bundle. Below the bundle, the allowance is consumed; above it, calls are billed at the per-tier rates above.

What’s bundled

Across all plans:

• Cluster federation (Daalu Edge) — unlimited clusters per tenant.
• The full feature set — alerts, automations, agents, Source of Truth, the Assistant, the AI Factory.
• Audit logs.
• Standard data retention — 90 days for events, 365 days for alerts and change proposals.

Extended retention is an Enterprise add-on.

What’s not bundled

Out of scope of Daalu’s pricing entirely:

• Your cloud bill. Daalu doesn’t proxy your cloud spend. You pay AWS / GCP / Azure directly; Daalu just shows the spend on its billing page.
• Your other vendors. PagerDuty, Slack, and the rest — same point.
• Your GPU hardware. You buy and run the card; Daalu routes to it for free.
• Storage of your own data. Daalu stores its own data (alerts, events, proposals). Your S3 bucket is yours.

Common questions

Can I bring my own LLM provider keys?

Yes — Enterprise. The router uses your keys for commercial inference, billed directly to you by the provider. Daalu’s inference metering then doesn’t apply to those calls.

Can I self-host Daalu?

The strictest data-residency needs are served today by Daalu Private, where your tenant’s data and agents run inside your own cluster. Fully operator-self-hosted deployment is a roadmap item — talk to Enterprise sales.

Is there a non-profit / academic rate?

Yes. Email [email protected] and tell us about your use.

What happens if I exceed a plan quota?

You stay in service. The overage lands on your next invoice at the published per-unit rates. We don’t gate your operations behind soft caps — only at hard tenant limits (e.g. an absolute database size in extreme cases).

What’s the billing cycle?

Monthly, in arrears. Invoices issue on the 1st for the previous month. Card payments by default; ACH / wire is available on Business and up.

Free trial

Every new tenant starts with a 14-day Team-tier trial — no card required. At day 14 the tenant converts to Starter automatically; we email a reminder at day 10.

Upgrade or downgrade any time. Downgrades take effect at the start of the next month.

Next: Chapter 47 — SKU breakdown