Weaviate
Open-source vector database with built-in ML vectorization modules (text2vec, multimodal) accessible via REST, GraphQL, and gRPC, available as self-hosted or managed Weaviate Cloud.
Score Breakdown
⚙ Agent Friendliness
🔒 Security
Self-hosted deployments default to no authentication and require explicit hardening; Weaviate Cloud enforces TLS and API key auth; OIDC support enables enterprise SSO integration.
⚡ Reliability
Best When
Building RAG or agent memory systems that need hybrid keyword+vector search and want built-in vectorization without managing a separate embedding service.
Avoid When
Your team is more comfortable with SQL interfaces or you need strict transactional consistency across vector and metadata updates.
Use Cases
- • Agent long-term memory store with semantic retrieval over episodic or procedural memory chunks
- • Hybrid search combining BM25 keyword and vector similarity in a single query for RAG pipelines
- • Multimodal agent memory that stores and retrieves both text and image embeddings in the same collection
- • Multi-tenant agent platforms where each user's data is isolated in a named tenant within a shared schema
- • Knowledge graph-style agent backends using Weaviate's cross-reference linking between object classes
Not For
- • Transactional workloads requiring ACID guarantees or row-level locking
- • Teams with no Kubernetes or Docker experience who need a zero-ops managed SQL alternative
- • Purely relational data with no vector search requirement
Interface
Authentication
Weaviate Cloud uses API key auth; self-hosted supports OIDC/OAuth2 with configurable providers; anonymous access can be enabled for local dev; read-only vs read-write key separation supported.
Pricing
Self-hosted is fully free and open-source under BSD-3; Weaviate Cloud managed service has sandbox tier without credit card.
Agent Metadata
Known Gotchas
- ⚠ GraphQL Get queries are the primary read path but agents often try REST first — the REST object endpoint does not support vector search and returns metadata only
- ⚠ Built-in vectorization modules (text2vec-openai, etc.) require API key configuration at collection creation time; adding them later requires recreating the collection
- ⚠ Multi-tenancy must be enabled at collection creation — it cannot be added to an existing collection and requires per-query tenant specification or queries silently return no results
- ⚠ The v4 Python client uses a different import structure and connection pattern than v3; mixing client versions against the same cluster produces confusing authentication errors
- ⚠ Hybrid search alpha parameter (0=keyword, 1=vector) defaults to 0.75 but is not documented prominently; agents using default hybrid may get unexpectedly keyword-dominated results
Alternatives
Full Evaluation Report
Detailed scoring breakdown, competitive positioning, security analysis, and improvement recommendations for Weaviate.
Scores are editorial opinions as of 2026-03-06.