Sticker price vs real cost
The per-seat headline is the easy part. The variable part, Cube Compute Units, is what surprises teams at renewal.
| What you see | What you actually pay |
|---|---|
| $40 / $80 per developer / month | Seat fees, plus consumption in Cube Compute Units (CCUs) |
| "Free tier available" | Free Cube Core to self-host (you pay infra); Cube Cloud free tier has capped AI tokens |
| "AI included" | Per-seat AI token grants equal to half the seat price; overage billed on-demand |
| Predictable monthly | CCU consumption scales with query load and pre-aggregation builds |
Cube Cloud tiers (2026)
| Plan | Seat | Consumption | Notes |
|---|---|---|---|
| Free | $0 | Capped monthly AI token allowance | Individual dev / evaluation |
| Starter | $40 / developer / mo | ~$0.10 / CCU, $99/mo minimum | On-demand billing via card |
| Premium | $80 / developer / mo | ~$0.25 / CCU, $10K/yr commit | 99.95% uptime SLA |
| Enterprise | Custom | ~$0.40 / CCU | Higher SLA, security, support, token packages |
Cube Core, the open-source engine, is Apache 2.0 and free to self-host. You trade the license cost for running and scaling the infrastructure yourself.
What actually drives your Cube bill
- Query volume. Every query against the metric API consumes compute. High-traffic embedded analytics can move CCU consumption fast.
- Pre-aggregations. Cube's speed comes from pre-aggregations, but building and refreshing them consumes CCUs and storage. The faster you want it, the more you build.
- Seats. Per-developer fees are linear, and AI token grants are pegged to seat price (half the seat cost per user).
- AI usage. Beyond the per-seat grant, AI requests bill as on-demand consumption or require pooled token packages on contracts.
Fix the Context, Not the Model. A consumption meter rewards a well-modeled, well-cached layer and punishes a sprawling one. The cheapest Cube bill and the most reliable answers come from the same thing: disciplined semantic modeling.
How Cube pricing compares
Cube's model is seats plus consumption. That contrasts with dbt's seats-plus-per-metric-query model and with warehouse-native tools that bill compute directly.
| Platform | Model | Entry |
|---|---|---|
| Cube Cloud | Per-seat + CCU consumption | Free tier; $40/dev/mo Starter |
| dbt Semantic Layer | Per-seat + ~$0.075/queried metric | Developer free; $100/user/mo Starter (teardown) |
| AtScale | Annual enterprise license | Custom, no public price (teardown) |
| Colrows | Priced on Semantic Assets, not seats or queries | Free ($0) + custom Enterprise |
Colrows deliberately does not meter on queries or seats. It prices on the number of governed Semantic Assets it manages, so cost tracks the size of your semantic layer, not how hard your agents hit it. For the architectural contrast, see Colrows vs Cube.
Frequently asked questions
How much does Cube Cloud cost?
Starter is $40 per developer per month and Premium is $80 per developer per month, plus consumption in Cube Compute Units (~$0.10/CCU Starter with a $99/month minimum, ~$0.25/CCU Premium with a $10K/year commit). Enterprise is custom, around $0.40/CCU. A free tier exists.
What is a Cube Compute Unit?
Cube Cloud's consumption metric. Your bill scales with CCUs consumed by query load and pre-aggregation builds, on top of seat fees.
Is Cube free?
Cube Core is open-source and free to self-host (you pay infrastructure). Cube Cloud has a free tier plus paid plans.
