Premium is custom-priced with $0.00975 per million dimensions, $0.31875 per GiB storage, $0.033 per GiB backup on Premium Dedicated. Premium is required for HIPAA, PrivateLink, customer-managed keys, or multi-region residency. Premium includes dedicated infrastructure, 99.95% uptime SLA, business-critical support, and a dedicated Technical Account Team.
Side-by-side: Weaviate Cloud tier comparison
| Plan | Per 1M dims | Storage/GiB | Backup/GiB | Min/mo | SLA |
|---|
| Sandbox | Free | Free | Free | $0 | None |
| Flex | $0.01668 | $0.255 | $0.0264 | $45 | 99.5% |
| Plus (Shared) | $0.0139 | $0.2125 | $0.022 | $280 | 99.9% |
| Premium (Dedicated) | $0.00975 | $0.31875 | $0.033 | Custom | 99.95% |
| Self-hosted | Free | Infrastructure | Infrastructure | ~$0 license | Self-managed |
Hidden costs and contract traps
Beyond the headline minimum, four cost factors determine Weaviate Cloud total cost of ownership. First, embedding services ($0.025-0.065 per 1M tokens depending on model) on top of storage costs. Second, Query Agent charges ($30/org/mo beyond the free tier) for AI-driven query features. Third, add-ons like enablement and training packages are separate costs. Fourth, Premium contract negotiations for custom SLAs and dedicated infrastructure.
The vector dimension calculation includes the replication factor, which means a 1M vector index with replication factor 3 bills as 3M vector dimensions. This is the most common pricing trap for teams that don't account for replication in their dimension count. Plan 3x the expected vector count if using default replication (Ranksquire, April 2026).
How Weaviate compares to Pinecone and Qdrant
Weaviate competes with Pinecone and Qdrant in the managed vector database market. Weaviate Flex at $45/mo minimum + per-dimension usage is similar to Pinecone Standard at $50/mo minimum + per-read/write usage. For typical RAG workloads, both platforms land in the $50-200/mo range. For large-scale workloads with high read volume, Pinecone's per-read pricing can spike while Weaviate's per-dimension pricing is more predictable.
Qdrant Cloud is quote-based without published rates, making direct comparison difficult. Self-hosting Qdrant or Weaviate on cloud infrastructure is often 60-80% cheaper at scale compared to managed cloud offerings (Ranksquire, April 2026).
Weaviate self-hosted: the free alternative
Self-hosted Weaviate under BSD-3 license is free and includes all features (vector search, hybrid search, multi-tenancy). You only pay for infrastructure costs. For teams with platform engineering capacity, self-hosted Weaviate on a $200-500/mo cloud VM can handle workloads that would cost $1,000-3,000/mo on Weaviate Cloud.
Self-hosting requires capacity for monitoring, scaling, patching, backups, and disaster recovery. Weaviate Cloud eliminates significant operational overhead. For most teams, Weaviate Cloud is the right balance of cost vs convenience. For teams with existing Kubernetes expertise and data residency requirements, self-hosting is often preferred.
What enterprise buyers should do next
Three actions for organizations evaluating Weaviate Cloud in 2026.
- Calculate accurate vector dimensions. Include the replication factor in your dimension count. A 1M vector index with replication factor 3 is 3M dimensions, not 1M.
- Compare Flex vs Plus for your workload. Flex is pay-as-you-go, Plus requires annual commitment. For variable workloads, Flex; for steady production, Plus.
- Add embedding costs to the model. Weaviate Embeddings $0.025-0.065 per 1M tokens depending on model is a separate line item from storage costs.
What to watch next
Three near-term datapoints. First, the Bundled Migration of legacy $25/mo Serverless tier customers to the new pricing model. Second, Weaviate's continued investment in AI-native features like Query Agent. Third, the competitive landscape - Pinecone and Qdrant are both investing in transparent per-dimension pricing with similar feature sets.