Pricing
Pricing Tiers
We offer Nucleus in four packages:
- Free Tier - Designed for academia and individuals who want to try out Nucleus with no commitments
- Team Tier - Designed for small ML teams who want to start optimizing their labeling spend
- Pro Tier - Designed for mid-sized ML teams who want to scale up their data curation
- Enterprise Tier - Designed for large ML teams to work with unlimited data
Pricing Breakdown
Feature Support
Consumption Categories
| Category | Description |
|---|---|
| Data Items Ingested | Total # of items ingested in Nucleus in a given month. Charged per 1000 items for all items above the allowed volume. |
| Items Scanned Over Queries | Total # items scanned while querying (including structured queries, NL queries, filter options & chart interactions). Duplicated queries run with 1 minute of each other are excluded. Charged per 1000 items scanned. |
| Internal Embeddings Generated | Total # of embeddings generated by Nucleus on the ingested data to enable NL search, similarity search, auto-tag and uniqueness sampling. Charged per 1000 embeddings generated. |
| Custom Embeddings Ingested | Total # of custom embeddings ingested into Nucleus by the user to enable NL search, similarity search, auto-tag and uniqueness sampling. Charged per 1000 embeddings ingested. |
| Items Processed for Autotag | Total # items processed during Autotag refinement. All datasets or slice tems are updated with a similairty score when an autotag is committed, this score is then used for querying. Charged per 1000 items processed. |
| Predictions Processed for IOU | Total # of predictions for which error metrics are calculated (IOU matches, class confusions & PR curve). Model error computation is triggered via API endpoint, and should be re-triggered whenever new predictions are uploaded to update evaluation metrics and charts. Charged per 1000 predictions processed. |
| Model Zoo Items Inferenced | Total # of items inferenced on Scale Model Zoo models. Scale Model Zoo is a growing library of pre-trained ML models that can be used to generate predictions for datasets in Nucleus. Charged per 1000 items inferenced. |
| Scenario Tests Evaluated | Total # of items evaluated per scenario test. Charged per 1000 items evaluated per test i.e. if you have 10 tests with 100 items each, you will be charged 10 times if you evaluate all of them. |
Updated about 2 years ago