Category · Developer & infrastructure
Vector search and retrieval databases for AI agents
Databases an agent can query for retrieval, with vector, keyword and hybrid search. Compared on retrieval accuracy, filters, how fast an update becomes searchable, p95 latency, setup effort and cost at a stated corpus size, with hosted and self-hosted costs reported apart.
Capability keys db.vector · db.hybrid · db.fulltext · db.filters · db.serverless · All tools
letme.dev/db.vector picks the top-graded tool in this list and says how to call it direct; calling through letme comes later.
The same listing from the live API. Graded results come first, then the official MCP registry when no graded-only filter is set.
https://www.anchorterminal.com/api/v1/searchFilters
| Compare | # | Tool | Category | Grade | Score | Agent rating | p95 | Context | Price / x402 | Auth | Where | Details |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 28 | Pinecone API + MCPPinecone · HTTP API | Retrieval | BB | 76.4 | 3.5 (8) | n/a | n/a | $20 / mo | API key | Hosted + local | ||
|
Managed, closed-source vector database with serverless indexes. Top strength OpenAPI files per API version, quarterly versions with at least 12 months of support each Top weakness Nine regional incidents between 9 July and 28 September 2026, four lasting more than an hour |
||||||||||||
| 30 | Supabase API + MCPSupabase · HTTP API | Databases | BB | 75.8 | 3.3 (8) | n/a | n/a | $25 / mo | OAuth or key | Hosted + local | ||
|
Hosted Postgres with an auto-generated REST API (PostgREST), GraphQL, auth, storage, realtime and Edge Functions, plus a Management API and an official MCP server. Top strength OAuth 2.1 with dynamic client registration, plus personal access tokens scoped to chosen projects and permissions Top weakness Several multi-hour platform incidents between 27 August and 30 September |
||||||||||||
| 37 | Qdrant API + MCPQdrant · HTTP API | Retrieval | BB | 75.3 | 3.6 (8) | n/a | n/a | Freemium | API key | Hosted + local | ||
|
Open-source vector database written in Rust, run yourself or on Qdrant Cloud. Top strength Database keys can be read-only, limited to chosen collections and expire after 90 days by default Top weakness The MCP server has 2 tools, can't manage collections or run filtered queries, and last released on 10 December 2025 |
||||||||||||
| 93 | Typesense API + MCPTypesense · HTTP API | Retrieval | BB | 70.9 | 3.0 (2) | n/a | n/a | $0.09 / GB | OAuth or key | Hosted | ||
|
Open-source (GPL-3.0) search engine with typo-tolerant keyword search, vector and hybrid search, filters, facets, geo search, JOINs and built-in embedding. Top strength Keyword, vector and hybrid search with filters, facets, sorting and geo in one request Top weakness Server v30.2 from April 2026 is still the newest stable release |
||||||||||||
| 131 | Weaviate API + MCPWeaviate · HTTP API | Retrieval | B | 68.2 | 3.5 (2) | n/a | n/a | $45 / mo | API key | Local | ||
|
Open-source vector database written in Go, run yourself or on Weaviate Cloud. Top strength Built-in MCP server with 4 typed tools, readOnly, destructive and idempotent hints on each, and write tools hidden until enabled Top weakness No public status page or incident history |
||||||||||||
| 174 | LanceDBLanceDB · SDK + MCP | Retrieval | B | 65.4 | 3.0 (2) | n/a | n/a | Freemium | None | Local | ||
|
Open-source embedded database on the Lance columnar format. Top strength Embedded, no server, and tables can live directly in S3, GCS or Azure storage Top weakness No self-serve hosted API. Enterprise is sold through a contact form |
||||||||||||
| 218 | Upstash Vector API + MCPUpstash · HTTP API | Retrieval | B | 62.4 | 3.0 (2) | n/a | n/a | $0.004 / 1k req | OAuth or key | Hosted | ||
|
Serverless vector database billed per request. Top strength $0.40 per 100,000 requests and a free plan with 10,000 queries and updates a day Top weakness The Vector changelog's last entry is August 2025 and the Python client last shipped in February 2025 |
||||||||||||
| 293 | Milvus and Zilliz Cloud API + MCPZilliz · HTTP API | Retrieval | C | 57.5 | 3.5 (2) | n/a | n/a | $0. / unit | API key | Hosted + local | ||
|
Milvus is the open-source (Apache-2.0, LF AI and Data) vector database, run embedded as Milvus Lite, standalone or distributed on Kubernetes. Top strength Apache-2.0 under the LF AI and Data Foundation, run embedded as Milvus Lite, standalone or distributed Top weakness The Zilliz MCP server's only PyPI release dates from June 2025 and predates an August 2026 fix for token forwarding |
||||||||||||
| 400 | Chroma API + MCPChroma · HTTP API | Retrieval | E | 45.4 | 3.0 (2) | n/a | n/a | $250 / mo | API key | Hosted + local | ||
|
Open-source retrieval database that runs in-process, as a local server or as Chroma Cloud, a serverless hosted service. Top strength Runs in-process, as a local server or serverless on Chroma Cloud with the same v2 API Top weakness CVE-2026-45829, CVSS 9.3, has no patched release more than four months after the advisory |
||||||||||||
| 439 | Epsilla Vector DatabaseEpsilla · HTTP API | Retrieval | F | 36 | 1.5 (2) | n/a | n/a | $29 / mo | API key | Hosted | ||
|
Open-source (GPL-3.0) vector database with dense and sparse vectors, built-in embedding, filters and hybrid search through client-side retrievers and an RRF reranker. Top strength GPL-3.0 server you can run from one Docker image Top weakness No tagged release since 29 November 2025 and no release notes since March 2025 |
||||||||||||
Nothing matches these filters. .
p95 latency and context cost come from our probes, which haven't run yet, so those columns start hidden. Grades run from AA to F, and agent-ready means BB or better. Filters, sorting and export run in your browser; the table is complete without JavaScript.
Indexed, not reviewed (6)
Listings sorted into this category from public catalogues (the official MCP registry, APIs.guru, the x402 Bazaar and OpenRouter), with facts and our own checks but no score, grade or rank. How the index works.
| Listing | Kind | What it does | Why it's here |
|---|---|---|---|
| ArcadeDB MCP Server arcadedb.com | MCP server | Built-in MCP server for ArcadeDB multi-model database (graph, document, vector, time-series) | vendor's own, widely used |
| brilliant.design MCP server brilliant.design | MCP server | Local MCP bridge: your coding agent designs on Brilliant's real vector canvas, fully on 127.0.0.1. | vendor's own |
| deepcontext wild-card.ai | MCP server | Advanced codebase indexing and semantic search MCP server | vendor's own |
| Magnificent Jobs magnificentjobs.com | MCP server | Semantic search over 3.5M+ live US job postings pulled straight from company ATSes. Read-only. | vendor's own |
| Picsha AI picsha.ai | MCP server | AI-native digital asset management: semantic search, generative image edits, and CDN delivery. | vendor's own |
| WalkMyPlan walkmyplan.com | MCP server | Describe a house or upload a DWG/vector-PDF plan and get a walkable, editable 3D model. | vendor's own |
How we test this category
The same corpus indexed and the same queries run on every database. We measure retrieval accuracy, filter behaviour, update delay, p95 latency, setup effort and cost at a stated corpus size, and report hosted services and self-hosted deployments separately. This test hasn't run yet, so Task success is pending and the grades here come from the categories assessed from public evidence.
How the ranking works
Every listing is scored 0 to 100 and given a grade from AA to F. In the October 2026 research run, 7 of the 9 weighted categories are scored from public evidence (status history, docs, pricing, terms, source and security pages) against a published checklist, with the reason and sources for every score on the listing. Performance and Task success wait for our probes and task suites, so their weight is shared across the rest until they run. Negative events deduct up to 15 points. Read the methodology.
For companies
Do agents find, use and choose your tools?
An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.