Pricing
Pricing
Three tiers. Public pricing. No embedding API fees — Dreambase is model-agnostic, you bring float32 arrays from whatever model your application already uses.
Starter
Evaluate Dreambase on a real workload. No time limit, no credit card.
$0 / month
- Up to 500K rows
- 500MB vector storage
- 10M query units/month
- Community support
- PostgreSQL wire protocol
- REST + Python SDK
Production
For LLM applications running at scale. Replaces two database subscriptions at roughly the combined price of both.
$149 / month
- Up to 50M rows
- 50GB vector storage
- Unlimited query units
- Email support (24h SLA)
- Custom embedding dimensions up to 4096
- Dedicated read replicas
Scale
Dedicated cluster, no resource sharing. For platforms with sustained write throughput or multi-region requirements.
$499 / month
- Unlimited rows
- 500GB vector storage
- Dedicated cluster
- Priority support + Slack channel
- Custom retention policies
- SLA: 99.9% uptime
Startup discount available — 50% off Production for 12 months for teams under 20 engineers. Email [email protected].
Common questions
One hybrid query — a statement that includes a
NEAR operator — counts as 1 unit. Pure SQL queries without a vector component count as 0.1 units. EXPLAIN HYBRID calls are not counted. There is no per-row storage read charge; query units cover compute only.
On Production, additional storage is $0.04/GB/month. On Scale, storage is negotiated as part of the contract.
Yes. Self-hosted Docker is available on all plans at no additional licensing cost. You pay only for support.
None. Dreambase is embedding-model agnostic. You generate embeddings in your application — using text-embedding-3-small, text-embedding-3-large, Cohere embed-v3, a fine-tuned sentence-transformer, or anything else that produces float32 arrays — and write them as a
VECTOR(dims) column. Dreambase does not call any external embedding API on your behalf.
Yes — email [email protected]. We offer 50% off Production for the first 12 months for companies under 20 engineers.