Snowflake Premium

Train at warehouse scale without exporting your data.

Connect your Snowflake account for query-in-place runs, higher API and training quotas, real-time Avalanche settlement, and a shared workspace from first experiment through production routing.

→Link repos, mint API keys, and track model versions in one place.
→Billing and access are tier-managed—no hand-wired metering per project.
→Snowflake query costs stay on your Snowflake bill—details in FAQ below.

Create workspace Whitepaper Sign in

FAQ

Metering, Snowflake, settlement, and platform basics for Premium teams.

Each prediction, inference, or model swap against Deadwood’s API counts as one call. Local includes 1,000 calls/month — fine for experiments and early integrations.

Can I change tiers monthly?

Yes. Downgrade anytime (refund prorated where applicable). Upgrade anytime.

What if I exceed my limits?

Local: requests are rejected until the next period. Premium: transparent overage pricing on your dashboard. Enterprise: no hard caps.

What if I need custom pricing?

Talk to our team — Enterprise is fully customizable for data residency, SLAs, and dedicated infrastructure.

Is there a free trial of Premium?

Yes. 14-day Premium trial — contact sales or start from the pricing CTAs when available.

A training run fine-tunes or trains a model from your configuration and data (local files or Snowflake-backed queries on Premium and Enterprise). Local includes one training run per month on local data up to 10K items.

On Premium, you connect your own Snowflake account. Deadwood queries data in place — models train without bulk-exporting your warehouse. On Enterprise, we can provision a dedicated Snowflake cluster with unlimited querying and enterprise residency controls.

Predictions and usage settle on Avalanche for fast, cheap, auditable records. Premium adds real-time settlement so balances and proofs stay in sync with production traffic.

On Premium, Snowflake query costs are billed by Snowflake to your account; Deadwood charges for training and API usage. On Enterprise, dedicated-cluster terms can include unlimited querying.

No. Deadwood trains against queries you authorize; massive datasets stay in your cloud.

Yes. Local workflows never require Snowflake. Snowflake is optional infrastructure for large-scale training.