DigitalOcean Launches Managed Weaviate in Public Preview
Provision a managed Weaviate cluster via DigitalOcean control panel or API; it runs Weaviate 1.37.1 with auto upgrades, RQ8 compression, and full API compatibility; no per‑query costs.
Provision a test cluster in public preview, verify API compatibility, and evaluate RQ8 compression impact on memory usage.
Summary
DigitalOcean has launched Managed Weaviate in public preview, offering a fully managed vector database that runs the open‑source Weaviate engine version 1.37.1. The service starts at $20/month with flat pricing, eliminating per‑query read unit charges and per‑dimension surcharges, and scales automatically on the same infrastructure as other DigitalOcean managed databases. Provisioning is done through the DigitalOcean control panel or API, and the cluster is fully compatible with Weaviate’s GraphQL, REST, and gRPC APIs on port 443, so existing client code requires no changes. DigitalOcean handles automated backups, security patching, version upgrades, high‑availability, and storage autoscaling, and the platform includes new preview features such as auto‑version upgrade, fork clusters, insights and logs, credential rotation, tags, and TLS by default.
Managed Weaviate also enables RQ8 (Rotational Quantization 8‑bit) compression by default, cutting RAM usage per vector by roughly four times while preserving recall. The service is open‑source, fully portable, and has no lock‑in, allowing customers to move clusters to other environments if desired. It integrates natively with DigitalOcean’s inference engine for embeddings, keeping the entire RAG pipeline co‑located without separate vendor relationships. The preview release is available immediately, and DigitalOcean plans to expand the feature set as it moves toward general availability.
Key changes
- Managed Weaviate launched on DigitalOcean, starting at $20/month with flat pricing and no per‑query or per‑dimension charges
- Runs Weaviate 1.37.1 with full API compatibility (GraphQL, REST, gRPC) on port 443, requiring no code changes
- DigitalOcean handles automated backups, security patching, version upgrades, high‑availability, and storage autoscaling
- New preview features include auto‑version upgrade, fork clusters, insights and logs, credential rotation, tags, and TLS by default
- RQ8 compression is enabled by default, reducing RAM usage per vector by ~4× while maintaining recall
- Integration with DigitalOcean Inference Engine for embeddings keeps the RAG stack co‑located and eliminates separate vendor relationships
- No per‑query read unit charges or per‑dimension surcharges, providing predictable scaling and cost