Spark is for first work
Build, query, export, and recover local files without card collection.
Plans
Start in Spark without a payment method. Upgrade when the model needs cloud save, larger graph limits, sharing, Model Discovery, or authenticated support.
Build, query, export, and recover local files without card collection.
Cloud documents, autosave, sharing, support, and larger limits unlock by account state.
Use Studio when unlisted read-only links and the largest self-service limits matter.
Spark
No cardEntry tier
$0 always available
An always-available entry workspace for learning, building, and testing small Bayesian networks.
Pro
Monthly billing
$29.99 charged monthly
Individual authoring with cloud saving, sampling, and invited sharing.
Premium
Billed every 6 months
$149.94 charged semi-annually
More room for larger models, more cloud files, and broader invited sharing.
Studio
PopularBilled annually
$239.88 charged annually
Largest self-serve limits, public read-only sharing, and best annual value.
Comparison
| Capability | Spark | Pro | Premium | Studio |
|---|---|---|---|---|
| Graph size | 10 nodes / 15 edges | 500 nodes / 1,000 edges | 1,500 nodes / 4,000 edges | 5,000 nodes / 15,000 edges |
| Sampling | 100 rows/run | 1,000 rows/run | 5,000 rows/run | 10,000 rows/run |
| Cloud documents | Local files only | 25 cloud files | 100 cloud files | 500 cloud files |
| Model Discovery | No Model Discovery | 25 Model Discovery requests/month | 100 Model Discovery requests/month | 500 Model Discovery requests/month |
| Sharing | No sharing | 3 invited recipients/share | 5 invited recipients/share | Public and invited read-only sharing |
| Support | Public support contact | Authenticated support tickets | Authenticated support tickets | Authenticated support tickets |
Spark does not collect a card. Paid upgrades use Stripe Checkout; Darkstar stores account, subscription, and billing-consent records, not card numbers.
Commercial SDK purchases are priced by language, platform, and license rights. Online plans cover the hosted workspace, not SDK redistribution or internal runtime rights.
Choose Spark for a first local model, Pro for individual cloud work, Premium for heavier saved analysis, and Studio when read-only sharing and larger self-service limits are part of the work.