{ "data": { "category": { "area": "developer", "capabilities": [ "db.vector", "db.hybrid", "db.fulltext", "db.filters", "db.serverless" ], "description": "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.", "indexed": [ { "jsonUrl": "https://www.anchorterminal.com/api/v1/tools/arcadedb-mcp-server.json", "kind": "mcp", "name": "ArcadeDB MCP Server", "slug": "arcadedb-mcp-server", "url": "https://www.anchorterminal.com/tools/arcadedb-mcp-server" }, { "jsonUrl": "https://www.anchorterminal.com/api/v1/tools/brilliant-mcp.json", "kind": "mcp", "name": "brilliant.design MCP server", "slug": "brilliant-mcp", "url": "https://www.anchorterminal.com/tools/brilliant-mcp" }, { "jsonUrl": "https://www.anchorterminal.com/api/v1/tools/wild-card-deepcontext.json", "kind": "mcp", "name": "deepcontext", "slug": "wild-card-deepcontext", "url": "https://www.anchorterminal.com/tools/wild-card-deepcontext" }, { "jsonUrl": "https://www.anchorterminal.com/api/v1/tools/magnificentjobs-jobs.json", "kind": "mcp", "name": "Magnificent Jobs", "slug": "magnificentjobs-jobs", "url": "https://www.anchorterminal.com/tools/magnificentjobs-jobs" }, { "jsonUrl": "https://www.anchorterminal.com/api/v1/tools/picsha.json", "kind": "mcp", "name": "Picsha AI", "slug": "picsha", "url": "https://www.anchorterminal.com/tools/picsha" }, { "jsonUrl": "https://www.anchorterminal.com/api/v1/tools/walkmyplan.json", "kind": "mcp", "name": "WalkMyPlan", "slug": "walkmyplan", "url": "https://www.anchorterminal.com/tools/walkmyplan" } ], "indexedCount": 6, "json": "https://www.anchorterminal.com/categories/vector-search.json", "name": "Retrieval \u0026 vector search", "slug": "vector-search", "test": "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.", "title": "Vector search and retrieval databases for AI agents", "toolCount": 14, "tools": [ "meilisearch", "pinecone", "supabase-mcp", "qdrant", "typesense", "vespa", "opensearch", "weaviate", "elasticsearch", "lancedb", "upstash-vector", "milvus-zilliz", "chroma", "epsilla" ], "url": "https://www.anchorterminal.com/categories/vector-search" }, "faq": [ { "answer": "Meilisearch has the highest benchmark score of the 14 ranked vector search and retrieval databases, 78.6 (A). Pinecone API + MCP is second with 76.2 (BB).", "question": "What are the highest-rated vector search and retrieval databases for AI agents?" }, { "answer": "6 of the 14 ranked here grade BB or better, the bar for agent-ready on the Anchor benchmark.", "question": "How many vector search and retrieval databases are agent-ready?" }, { "answer": "None of the ranked listings here accepts x402 for its main call yet.", "question": "Which vector search and retrieval databases accept x402 payments?" }, { "answer": "By published paid prices, Milvus and Zilliz Cloud API + MCP, at $0.025 per GB, the lowest of the 10 listings here with a paid price in this unit (free allowances aside). Plans, volume tiers and free allowances change the sum, so check the listing's price table.", "question": "Which of these vector search and retrieval databases is cheapest?" }, { "answer": "By the Anchor benchmark score out of 100, a weighted mean of the scored categories minus deductions for negative events, from public evidence re-checked as vendors change. Listings cannot pay for a place. The latest assessment behind this page is from 9 October 2026.", "question": "How is this list ranked?" } ], "howToChoose": [ { "label": "Hosted and self-hosted costs", "detail": "Compare hosted and self-hosted costs at the same corpus size and vector dimension, since a unit price means little until stored and query volumes are fixed." }, { "label": "Update delay before searchable", "detail": "Check how long after a write an update becomes searchable, because an agent that stores a fact and immediately queries for it will miss the fact if indexing lags." }, { "label": "Filter order and result counts", "detail": "Check whether metadata filters are applied before or after ranking, since post-filtering can return fewer results than requested and silently weaken retrieval accuracy." }, { "label": "Retrieval accuracy on your corpus", "detail": "Check whether retrieval accuracy was measured on your own corpus and queries, since a score on another corpus says little about how well results will rank for your agent." } ], "picks": [ { "also": { "name": "Pinecone API + MCP", "slug": "pinecone", "why": "BB, 76.2/100" }, "name": "Meilisearch", "need": "Highest score overall", "slug": "meilisearch", "why": "A, 78.6/100 on the benchmark" }, { "name": "Pinecone API + MCP", "need": "Schema \u0026 documentation", "slug": "pinecone", "why": "94/100 on schema \u0026 documentation, against 93 for the overall leader" }, { "name": "Pinecone API + MCP", "need": "Agent ergonomics", "slug": "pinecone", "why": "93/100 on agent ergonomics, against 87 for the overall leader" }, { "name": "Supabase API + MCP", "need": "Security \u0026 auth", "slug": "supabase-mcp", "why": "84/100 on security \u0026 auth, against 67 for the overall leader" }, { "name": "Supabase API + MCP", "need": "Transparency \u0026 trust", "slug": "supabase-mcp", "why": "90/100 on transparency \u0026 trust, against 74 for the overall leader" }, { "also": { "name": "Weaviate API + MCP", "slug": "weaviate", "why": "$0.029 per GB" }, "name": "Milvus and Zilliz Cloud API + MCP", "need": "Lowest paid price per GB", "slug": "milvus-zilliz", "why": "$0.025 per GB, the lowest of the 10 listings here with a paid price in this unit (free allowances aside)" }, { "name": "Pinecone API + MCP", "need": "A hosted MCP endpoint", "slug": "pinecone", "why": "remote MCP server, nothing to install" }, { "name": "Qdrant API + MCP", "need": "The review panel's favourite", "slug": "qdrant", "why": "3.6/5 from 8 panel reviews" } ], "ranked": 14, "shortlist": [ { "bestFor": "Retrieval where typo-tolerant keyword relevance, filters and facets matter as much as vectors, with embeddings generated by the engine.", "grade": "A", "name": "Meilisearch", "position": 1, "price": "$30 / mo", "score": 78.6, "slug": "meilisearch", "strengths": [ "Keyword, vector and hybrid search in one request, with `semanticRatio`, filters, facets, sorting and federated multi-index search", "OpenAPI 3.1 file for v1.54.3 with 143 operations, each described, checked daily in CI and attached to every release", "API keys limited to named actions and indexes with an expiry, plus tenant tokens that embed per-user search filters" ], "url": "https://www.anchorterminal.com/tools/meilisearch", "verdict": "Keyword, vector and hybrid search share one request, behind keys scoped by action and index and a published OpenAPI 3.1 file with 143 operations. The built-in MCP server is experimental and read-only, and Cloud access starts with a browser sign-up and a 14-day trial.", "weaknesses": [ "The `/mcp` route is an experimental feature that must be switched on, and its four tools carry no read-only annotations", "No per-plan request rate limit is published for Cloud, although the monitoring docs list rate limiting as a cause of failed calls", "The 99.9 per cent uptime SLA applies to Enterprise plans only" ], "where": "local", "x402": "no" }, { "bestFor": "Teams that want a hosted index with nothing to run, hosted embedding and reranking, and a versioned API agents can rely on.", "grade": "BB", "name": "Pinecone API + MCP", "position": 2, "price": "$20 / mo", "score": 76.2, "slug": "pinecone", "strengths": [ "OpenAPI files per API version, quarterly versions with at least 12 months of support each", "Published per-namespace and per-index rate limits, 429 responses and backoff guidance", "MCP tool descriptions say when a tool will fail, and every tool carries readOnly or destructive hints" ], "url": "https://www.anchorterminal.com/tools/pinecone", "verdict": "OpenAPI files per API version, quarterly versions with at least 12 months of support each. Nine regional incidents between 9 July and 28 September 2026, four lasting more than an hour.", "weaknesses": [ "Nine regional incidents between 9 July and 28 September 2026, four lasting more than an hour", "Every MCP database tool asks the model for its provider and model name for analytics, which the README doesn't mention", "The Developer MCP server only works with integrated-embedding indexes and has no read-only mode" ], "where": "both", "x402": "no" }, { "bestFor": "Agents building on or administering Supabase projects, and for retrieval that wants vectors, full-text and SQL filters in one database.", "grade": "BB", "name": "Supabase API + MCP", "position": 3, "price": "$25 / mo", "score": 75.6, "slug": "supabase-mcp", "strengths": [ "OAuth 2.1 with dynamic client registration, plus personal access tokens scoped to chosen projects and permissions", "`read_only`, `project_ref` and `features` cut the server from 34 tools to as few as 6 and run SQL as a read-only role", "Destructive SQL and cost-bearing creates ask for confirmation through elicitation since v0.13.0" ], "url": "https://www.anchorterminal.com/tools/supabase-mcp", "verdict": "OAuth 2.1 with dynamic client registration, plus personal access tokens scoped to chosen projects and permissions. Several multi-hour platform incidents between 27 August and 30 September.", "weaknesses": [ "Several multi-hour platform incidents between 27 August and 30 September", "Read-write with seven feature groups is the default, and the agent plugin has no read-only option", "Untrusted data in tables can still steer an agent that reads it, as Supabase says itself" ], "where": "both", "x402": "no" }, { "bestFor": "Filtered vector search and multi-tenant retrieval where the agent can call REST, and for teams that want the same engine self-hosted and managed.", "grade": "BB", "name": "Qdrant API + MCP", "position": 4, "price": "Freemium", "score": 75.3, "slug": "qdrant", "strengths": [ "Database keys can be read-only, limited to chosen collections and expire after 90 days by default", "429 responses carry `Retry-After` in seconds, and `wait=true` makes writes read-your-own-write", "Published SLAs from 99.5% on Free and Standard to 99.95% with multi-AZ" ], "url": "https://www.anchorterminal.com/tools/qdrant", "verdict": "Database keys can be read-only, limited to chosen collections and expire after 90 days by default. The MCP server has 2 tools, can't manage collections or run filtered queries, and last released on 10 December 2025.", "weaknesses": [ "The MCP server has 2 tools, can't manage collections or run filtered queries, and last released on 10 December 2025", "No per-unit rate card for Qdrant Cloud, only a calculator", "Self-hosted builds send usage statistics by default until you opt out" ], "where": "both", "x402": "no" }, { "bestFor": "Retrieval where keyword relevance, typo tolerance, filters and facets matter as much as vectors, and for teams that want a hosted MCP server with tight permissions.", "grade": "BB", "name": "Typesense API + MCP", "position": 5, "price": "$0.09 / GB", "score": 70.7, "slug": "typesense", "strengths": [ "Keyword, vector and hybrid search with filters, facets, sorting and geo in one request", "Hosted MCP server with OAuth consent presets, server-held data keys and destructive-tool annotations", "No status notices from July to October 2026 across 21 regions" ], "url": "https://www.anchorterminal.com/tools/typesense", "verdict": "Keyword, vector and hybrid search with filters, facets, sorting and geo in one request. Server v30.2 from April 2026 is still the newest stable release.", "weaknesses": [ "Server v30.2 from April 2026 is still the newest stable release", "The authentication docs show the API key as a URL query parameter", "No SLA found, and SOC 2 reports need a paid support plan" ], "where": "hosted", "x402": "no" }, { "bestFor": "Teams that need hybrid retrieval with custom ranking at scale and can maintain a schema and application package.", "grade": "BB", "name": "Vespa", "position": 6, "price": "$0.05 / vCPU-hr", "score": 70, "slug": "vespa", "strengths": [ "Data-plane clients use mTLS certificates or tokens, each limited to `read`, `write` or both, with token versions, expiry and revocation", "Unit prices are public, from $0.05 a vCPU-hour on Startup, and the trial gives $300 of credits with no card", "Every docs page has a Markdown twin, with `llms.txt` and `llms-full.txt` on docs.vespa.ai" ], "url": "https://www.anchorterminal.com/tools/vespa", "verdict": "Vespa Cloud publishes hourly unit prices, a 99.5% uptime agreement for paying customers and a trial with $300 of credits and no card. Every call needs an application package deployed first, no OpenAPI file was found in the reviewed documentation, and Vespa.ai says it has no official public MCP server.", "weaknesses": [ "No OpenAPI file was found in the docs index. The APIs are described in reference pages only", "Vespa.ai states it has no official public MCP server. An MCP endpoint must be added to the user's own application", "No call works until a schema and `services.xml` are deployed as an application package" ], "where": "local", "x402": "no" }, { "bestFor": "A team that already runs OpenSearch for logs or text search and wants vectors and hybrid search in the same cluster, with fine-grained access control.", "grade": "B", "name": "OpenSearch", "position": 7, "price": "Free · OSS", "score": 68, "slug": "opensearch", "strengths": [ "Apache-2.0 for the engine, the plugins, the clients and the MCP server, under a foundation hosted by The Linux Foundation", "An OpenAPI 3.1 description of the REST API is published as one file, with about 717 operations in 36 namespace files in the source repository", "API keys since 3.7 carry cluster and index permissions, expire within 90 days at most, and are revoked on every node at once" ], "url": "https://www.anchorterminal.com/tools/opensearch", "verdict": "OpenSearch 3.9.0 runs vector, text and hybrid search under Apache-2.0, with an OpenAPI 3.1 description of its REST API and API keys limited by permission and index pattern. The owner runs the cluster, no status page or SLA comes with the software, and no tool in the MCP server sets a read-only or destructive annotation.", "weaknesses": [ "No hosted service from the foundation. The owner installs, secures and runs the cluster, and no status page or SLA exists for the software", "No `llms.txt` on docs.opensearch.org (HTTP 404), and doc pages carry about 160 KB of navigation text each", "The MCP server allows writes by default (`OPENSEARCH_SETTINGS_ALLOW_WRITE` is `true`) and no tool sets `readOnlyHint` or `destructiveHint`" ], "where": "local", "x402": "no" }, { "bestFor": "Hybrid keyword and vector retrieval with filters and multi-tenancy, self-hosted or managed, and for agents that work over MCP.", "grade": "B", "name": "Weaviate API + MCP", "position": 8, "price": "$45 / mo", "score": 67.9, "slug": "weaviate", "strengths": [ "Built-in MCP server with 4 typed tools, readOnly, destructive and idempotent hints on each, and write tools hidden until enabled", "Cloud API keys carry an RBAC role and can be rotated, and RBAC decisions go to an audit log", "Per-unit cloud prices published, and a free cluster with no card" ], "url": "https://www.anchorterminal.com/tools/weaviate", "verdict": "Built-in MCP server with 4 typed tools, readOnly, destructive and idempotent hints on each, and write tools hidden until enabled. No public status page or incident history.", "weaknesses": [ "No public status page or incident history", "No request limits or 429 guidance published for the database API", "Self-hosted telemetry is on by default, and the docs say it may be shared with third parties to find customers" ], "where": "local", "x402": "no" }, { "bestFor": "Retrieval that needs keyword relevance, filters and vectors in one query, and teams already running Elastic for logs or security.", "grade": "B", "name": "Elasticsearch", "position": 9, "price": "$0.09 / unit", "score": 67.7, "slug": "elasticsearch", "strengths": [ "OpenAPI 3.0.3 file for the Serverless API with 311 operations, plus llms.txt and a Markdown twin of every docs page", "API keys take role descriptors that limit indices and privileges, with an expiry, and the MCP server accepts OAuth 2.1 on Serverless", "Agent Builder MCP server is generally available on Serverless and on Elastic Stack 9.3 and later" ], "url": "https://www.anchorterminal.com/tools/elasticsearch", "verdict": "Keyword, dense and sparse vector, and hybrid search with rank fusion sit behind one OpenAPI-described REST API, with API keys scoped to indices and privileges. The status page lists more than 20 incidents marked major since July 2026, and Elasticsearch 9.5.0 returned incorrect results for some queries without raising an error.", "weaknesses": [ "More than 20 status notices marked major between 10 July and 8 October 2026, three of them regional service incidents", "Elasticsearch 9.5.0 (29 July 2026) returned incorrect results for some `must_not` queries with no error, fixed in 9.5.1 and 9.5.2", "No request rate limit with numbers was found in the reviewed documentation" ], "where": "local", "x402": "no" }, { "bestFor": "Agents whose code can import a library and keep retrieval data in object storage, from notebooks to multimodal datasets.", "grade": "B", "name": "LanceDB", "position": 10, "price": "Freemium", "score": 65.4, "slug": "lancedb", "strengths": [ "Embedded, no server, and tables can live directly in S3, GCS or Azure storage", "Four releases since late July, v0.39.0 on 17 September 2026, with passing CI on main", "Release notes with Breaking Changes and Deprecations sections, and minor-version bumps for breaking changes" ], "url": "https://www.anchorterminal.com/tools/lancedb", "verdict": "Embedded, no server, and tables can live directly in S3, GCS or Azure storage. No self-serve hosted API. Enterprise is sold through a contact form.", "weaknesses": [ "No self-serve hosted API. Enterprise is sold through a contact form", "No official MCP server, only a 3-tool reference example", "The docs' OpenAPI link for the Enterprise REST API returns 404" ], "where": "local", "x402": "no" } ], "updated": "2026-10-09" }, "kind": "anchor.page", "links": { "api": "https://www.anchorterminal.com/api/v1/index.json", "html": "https://www.anchorterminal.com/best/vector-search/", "json": "https://www.anchorterminal.com/best/vector-search/index.json", "llms": "https://www.anchorterminal.com/llms.txt", "markdown": "https://www.anchorterminal.com/best/vector-search/index.md", "slim": "https://www.anchorterminal.com/best/vector-search/index.min.md" }, "markdown": "The 10 highest-scoring of 14 vector search and retrieval databases on the Anchor benchmark, with a pick for each need and where each one falls short. Scores come from public evidence, re-checked as vendors change.\n\n- Ranked: 14 · agent-ready (BB or better): 6 · accept x402: 0 · hosted endpoints: 8\n- Full ranked table: https://www.anchorterminal.com/categories/vector-search.md\n- Head-to-head comparisons: https://www.anchorterminal.com/compare/vector-search/index.md (85)\n- Methodology: https://www.anchorterminal.com/benchmark/index.md\n\n## The shortlist\n\n| # | Tool | Grade | Score | Best for | Price | Where |\n| --- | --- | --- | --- | --- | --- | --- |\n| 1 | [Meilisearch](https://www.anchorterminal.com/tools/meilisearch.md) | A | 78.6 | Retrieval where typo-tolerant keyword relevance, filters and facets matter as much as vectors, with embeddings generated by the engine. | $30 / mo | local |\n| 2 | [Pinecone API + MCP](https://www.anchorterminal.com/tools/pinecone.md) | BB | 76.2 | Teams that want a hosted index with nothing to run, hosted embedding and reranking, and a versioned API agents can rely on. | $20 / mo | hosted and local |\n| 3 | [Supabase API + MCP](https://www.anchorterminal.com/tools/supabase-mcp.md) | BB | 75.6 | Agents building on or administering Supabase projects, and for retrieval that wants vectors, full-text and SQL filters in one database. | $25 / mo | hosted and local |\n| 4 | [Qdrant API + MCP](https://www.anchorterminal.com/tools/qdrant.md) | BB | 75.3 | Filtered vector search and multi-tenant retrieval where the agent can call REST, and for teams that want the same engine self-hosted and managed. | Freemium | hosted and local |\n| 5 | [Typesense API + MCP](https://www.anchorterminal.com/tools/typesense.md) | BB | 70.7 | Retrieval where keyword relevance, typo tolerance, filters and facets matter as much as vectors, and for teams that want a hosted MCP server with tight permissions. | $0.09 / GB | hosted |\n| 6 | [Vespa](https://www.anchorterminal.com/tools/vespa.md) | BB | 70 | Teams that need hybrid retrieval with custom ranking at scale and can maintain a schema and application package. | $0.05 / vCPU-hr | local |\n| 7 | [OpenSearch](https://www.anchorterminal.com/tools/opensearch.md) | B | 68 | A team that already runs OpenSearch for logs or text search and wants vectors and hybrid search in the same cluster, with fine-grained access control. | Free · OSS | local |\n| 8 | [Weaviate API + MCP](https://www.anchorterminal.com/tools/weaviate.md) | B | 67.9 | Hybrid keyword and vector retrieval with filters and multi-tenancy, self-hosted or managed, and for agents that work over MCP. | $45 / mo | local |\n| 9 | [Elasticsearch](https://www.anchorterminal.com/tools/elasticsearch.md) | B | 67.7 | Retrieval that needs keyword relevance, filters and vectors in one query, and teams already running Elastic for logs or security. | $0.09 / unit | local |\n| 10 | [LanceDB](https://www.anchorterminal.com/tools/lancedb.md) | B | 65.4 | Agents whose code can import a library and keep retrieval data in object storage, from notebooks to multimodal datasets. | Freemium | local |\n\n## Picks by need\n\n- Highest score overall: [Meilisearch](https://www.anchorterminal.com/tools/meilisearch.md), A, 78.6/100 on the benchmark. Also [Pinecone API + MCP](https://www.anchorterminal.com/tools/pinecone.md), BB, 76.2/100.\n- Schema \u0026 documentation: [Pinecone API + MCP](https://www.anchorterminal.com/tools/pinecone.md), 94/100 on schema \u0026 documentation, against 93 for the overall leader.\n- Agent ergonomics: [Pinecone API + MCP](https://www.anchorterminal.com/tools/pinecone.md), 93/100 on agent ergonomics, against 87 for the overall leader.\n- Security \u0026 auth: [Supabase API + MCP](https://www.anchorterminal.com/tools/supabase-mcp.md), 84/100 on security \u0026 auth, against 67 for the overall leader.\n- Transparency \u0026 trust: [Supabase API + MCP](https://www.anchorterminal.com/tools/supabase-mcp.md), 90/100 on transparency \u0026 trust, against 74 for the overall leader.\n- Lowest paid price per GB: [Milvus and Zilliz Cloud API + MCP](https://www.anchorterminal.com/tools/milvus-zilliz.md), $0.025 per GB, the lowest of the 10 listings here with a paid price in this unit (free allowances aside). Also [Weaviate API + MCP](https://www.anchorterminal.com/tools/weaviate.md), $0.029 per GB.\n- A hosted MCP endpoint: [Pinecone API + MCP](https://www.anchorterminal.com/tools/pinecone.md), remote MCP server, nothing to install.\n- The review panel's favourite: [Qdrant API + MCP](https://www.anchorterminal.com/tools/qdrant.md), 3.6/5 from 8 panel reviews.\n\n## How to choose\n\n- Hosted and self-hosted costs: Compare hosted and self-hosted costs at the same corpus size and vector dimension, since a unit price means little until stored and query volumes are fixed.\n- Update delay before searchable: Check how long after a write an update becomes searchable, because an agent that stores a fact and immediately queries for it will miss the fact if indexing lags.\n- Filter order and result counts: Check whether metadata filters are applied before or after ranking, since post-filtering can return fewer results than requested and silently weaken retrieval accuracy.\n- Retrieval accuracy on your corpus: Check whether retrieval accuracy was measured on your own corpus and queries, since a score on another corpus says little about how well results will rank for your agent.\n\n- How the benchmark tests 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.\n\n## Each one in detail\n\n### 1. Meilisearch, A 78.6/100\n\nOpen-source search engine from Meili SAS in Paris with typo-tolerant keyword search, vector and hybrid search, filters and facets. It runs self-hosted or as Meilisearch Cloud, with a REST API, SDKs and an experimental built-in MCP route.\n\n- Verdict: Keyword, vector and hybrid search share one request, behind keys scoped by action and index and a published OpenAPI 3.1 file with 143 operations. The built-in MCP server is experimental and read-only, and Cloud access starts with a browser sign-up and a 14-day trial.\n- Choose it for: Retrieval where typo-tolerant keyword relevance, filters and facets matter as much as vectors, with embeddings generated by the engine.\n- Strength: Keyword, vector and hybrid search in one request, with `semanticRatio`, filters, facets, sorting and federated multi-index search\n- Strength: OpenAPI 3.1 file for v1.54.3 with 143 operations, each described, checked daily in CI and attached to every release\n- Strength: API keys limited to named actions and indexes with an expiry, plus tenant tokens that embed per-user search filters\n- Weakness: The `/mcp` route is an experimental feature that must be switched on, and its four tools carry no read-only annotations\n- Weakness: No per-plan request rate limit is published for Cloud, although the monitoring docs list rate limiting as a cause of failed calls\n- Weakness: The 99.9 per cent uptime SLA applies to Enterprise plans only\n- Price: $30 / mo · Auth: API key · x402: no · Where: local\n- Full assessment: https://www.anchorterminal.com/tools/meilisearch.md\n\n### 2. Pinecone API + MCP, BB 76.2/100\n\nManaged, closed-source vector database with serverless indexes.\n\n- Verdict: OpenAPI files per API version, quarterly versions with at least 12 months of support each. Nine regional incidents between 9 July and 28 September 2026, four lasting more than an hour.\n- Choose it for: Teams that want a hosted index with nothing to run, hosted embedding and reranking, and a versioned API agents can rely on.\n- Strength: OpenAPI files per API version, quarterly versions with at least 12 months of support each\n- Strength: Published per-namespace and per-index rate limits, 429 responses and backoff guidance\n- Strength: MCP tool descriptions say when a tool will fail, and every tool carries readOnly or destructive hints\n- Weakness: Nine regional incidents between 9 July and 28 September 2026, four lasting more than an hour\n- Weakness: Every MCP database tool asks the model for its provider and model name for analytics, which the README doesn't mention\n- Weakness: The Developer MCP server only works with integrated-embedding indexes and has no read-only mode\n- Price: $20 / mo · Auth: API key · x402: no · Where: hosted and local\n- Full assessment: https://www.anchorterminal.com/tools/pinecone.md\n- Against #1: https://www.anchorterminal.com/compare/meilisearch-vs-pinecone.md\n\n### 3. Supabase API + MCP, BB 75.6/100\n\nHosted Postgres with an auto-generated REST API (PostgREST), GraphQL, auth, storage, realtime and Edge Functions, plus a Management API and an official MCP server.\n\n- Verdict: OAuth 2.1 with dynamic client registration, plus personal access tokens scoped to chosen projects and permissions. Several multi-hour platform incidents between 27 August and 30 September.\n- Choose it for: Agents building on or administering Supabase projects, and for retrieval that wants vectors, full-text and SQL filters in one database.\n- Strength: OAuth 2.1 with dynamic client registration, plus personal access tokens scoped to chosen projects and permissions\n- Strength: `read_only`, `project_ref` and `features` cut the server from 34 tools to as few as 6 and run SQL as a read-only role\n- Strength: Destructive SQL and cost-bearing creates ask for confirmation through elicitation since v0.13.0\n- Weakness: Several multi-hour platform incidents between 27 August and 30 September\n- Weakness: Read-write with seven feature groups is the default, and the agent plugin has no read-only option\n- Weakness: Untrusted data in tables can still steer an agent that reads it, as Supabase says itself\n- Price: $25 / mo · Auth: OAuth or key · x402: no · Where: hosted and local\n- Full assessment: https://www.anchorterminal.com/tools/supabase-mcp.md\n\n### 4. Qdrant API + MCP, BB 75.3/100\n\nOpen-source vector database written in Rust, run yourself or on Qdrant Cloud.\n\n- Verdict: Database keys can be read-only, limited to chosen collections and expire after 90 days by default. The MCP server has 2 tools, can't manage collections or run filtered queries, and last released on 10 December 2025.\n- Choose it for: Filtered vector search and multi-tenant retrieval where the agent can call REST, and for teams that want the same engine self-hosted and managed.\n- Strength: Database keys can be read-only, limited to chosen collections and expire after 90 days by default\n- Strength: 429 responses carry `Retry-After` in seconds, and `wait=true` makes writes read-your-own-write\n- Strength: Published SLAs from 99.5% on Free and Standard to 99.95% with multi-AZ\n- Weakness: The MCP server has 2 tools, can't manage collections or run filtered queries, and last released on 10 December 2025\n- Weakness: No per-unit rate card for Qdrant Cloud, only a calculator\n- Weakness: Self-hosted builds send usage statistics by default until you opt out\n- Price: Freemium · Auth: API key · x402: no · Where: hosted and local\n- Full assessment: https://www.anchorterminal.com/tools/qdrant.md\n- Against #1: https://www.anchorterminal.com/compare/meilisearch-vs-qdrant.md\n\n### 5. Typesense API + MCP, BB 70.7/100\n\nOpen-source (GPL-3.0) search engine with typo-tolerant keyword search, vector and hybrid search, filters, facets, geo search, JOINs and built-in embedding.\n\n- Verdict: Keyword, vector and hybrid search with filters, facets, sorting and geo in one request. Server v30.2 from April 2026 is still the newest stable release.\n- Choose it for: Retrieval where keyword relevance, typo tolerance, filters and facets matter as much as vectors, and for teams that want a hosted MCP server with tight permissions.\n- Strength: Keyword, vector and hybrid search with filters, facets, sorting and geo in one request\n- Strength: Hosted MCP server with OAuth consent presets, server-held data keys and destructive-tool annotations\n- Strength: No status notices from July to October 2026 across 21 regions\n- Weakness: Server v30.2 from April 2026 is still the newest stable release\n- Weakness: The authentication docs show the API key as a URL query parameter\n- Weakness: No SLA found, and SOC 2 reports need a paid support plan\n- Price: $0.09 / GB · Auth: OAuth or key · x402: no · Where: hosted\n- Full assessment: https://www.anchorterminal.com/tools/typesense.md\n- Against #1: https://www.anchorterminal.com/compare/meilisearch-vs-typesense.md\n\n### 6. Vespa, BB 70/100\n\nVespa is an open-source search engine from Vespa.ai AS for vector, text and hybrid retrieval with ranking, run self-hosted or on Vespa Cloud. Agents reach it through HTTP query and document APIs, the Vespa CLI and a Python client.\n\n- Verdict: Vespa Cloud publishes hourly unit prices, a 99.5% uptime agreement for paying customers and a trial with $300 of credits and no card. Every call needs an application package deployed first, no OpenAPI file was found in the reviewed documentation, and Vespa.ai says it has no official public MCP server.\n- Choose it for: Teams that need hybrid retrieval with custom ranking at scale and can maintain a schema and application package.\n- Strength: Data-plane clients use mTLS certificates or tokens, each limited to `read`, `write` or both, with token versions, expiry and revocation\n- Strength: Unit prices are public, from $0.05 a vCPU-hour on Startup, and the trial gives $300 of credits with no card\n- Strength: Every docs page has a Markdown twin, with `llms.txt` and `llms-full.txt` on docs.vespa.ai\n- Weakness: No OpenAPI file was found in the docs index. The APIs are described in reference pages only\n- Weakness: Vespa.ai states it has no official public MCP server. An MCP endpoint must be added to the user's own application\n- Weakness: No call works until a schema and `services.xml` are deployed as an application package\n- Price: $0.05 / vCPU-hr · Auth: OAuth or key · x402: no · Where: local\n- Full assessment: https://www.anchorterminal.com/tools/vespa.md\n- Against #1: https://www.anchorterminal.com/compare/meilisearch-vs-vespa.md\n\n### 7. OpenSearch, B 68/100\n\nOpenSearch is an open-source search and analytics engine with vector, text and hybrid search, run by its owner. Agents reach it through a REST API, official clients, an OpenSearch MCP Server and an MCP endpoint inside the cluster.\n\n- Verdict: OpenSearch 3.9.0 runs vector, text and hybrid search under Apache-2.0, with an OpenAPI 3.1 description of its REST API and API keys limited by permission and index pattern. The owner runs the cluster, no status page or SLA comes with the software, and no tool in the MCP server sets a read-only or destructive annotation.\n- Choose it for: A team that already runs OpenSearch for logs or text search and wants vectors and hybrid search in the same cluster, with fine-grained access control.\n- Strength: Apache-2.0 for the engine, the plugins, the clients and the MCP server, under a foundation hosted by The Linux Foundation\n- Strength: An OpenAPI 3.1 description of the REST API is published as one file, with about 717 operations in 36 namespace files in the source repository\n- Strength: API keys since 3.7 carry cluster and index permissions, expire within 90 days at most, and are revoked on every node at once\n- Weakness: No hosted service from the foundation. The owner installs, secures and runs the cluster, and no status page or SLA exists for the software\n- Weakness: No `llms.txt` on docs.opensearch.org (HTTP 404), and doc pages carry about 160 KB of navigation text each\n- Weakness: The MCP server allows writes by default (`OPENSEARCH_SETTINGS_ALLOW_WRITE` is `true`) and no tool sets `readOnlyHint` or `destructiveHint`\n- Price: Free · OSS · Auth: OAuth or key · x402: no · Where: local\n- Full assessment: https://www.anchorterminal.com/tools/opensearch.md\n- Against #1: https://www.anchorterminal.com/compare/meilisearch-vs-opensearch.md\n\n### 8. Weaviate API + MCP, B 67.9/100\n\nOpen-source vector database written in Go, run yourself or on Weaviate Cloud.\n\n- Verdict: Built-in MCP server with 4 typed tools, readOnly, destructive and idempotent hints on each, and write tools hidden until enabled. No public status page or incident history.\n- Choose it for: Hybrid keyword and vector retrieval with filters and multi-tenancy, self-hosted or managed, and for agents that work over MCP.\n- Strength: Built-in MCP server with 4 typed tools, readOnly, destructive and idempotent hints on each, and write tools hidden until enabled\n- Strength: Cloud API keys carry an RBAC role and can be rotated, and RBAC decisions go to an audit log\n- Strength: Per-unit cloud prices published, and a free cluster with no card\n- Weakness: No public status page or incident history\n- Weakness: No request limits or 429 guidance published for the database API\n- Weakness: Self-hosted telemetry is on by default, and the docs say it may be shared with third parties to find customers\n- Price: $45 / mo · Auth: API key · x402: no · Where: local\n- Full assessment: https://www.anchorterminal.com/tools/weaviate.md\n- Against #1: https://www.anchorterminal.com/compare/meilisearch-vs-weaviate.md\n\n### 9. Elasticsearch, B 67.7/100\n\nElasticsearch is Elastic's search engine and vector database, with keyword, vector and hybrid search over a REST API. Agents reach it through that API or the Agent Builder MCP server on Elastic Cloud and Elastic Stack 9.2 or later.\n\n- Verdict: Keyword, dense and sparse vector, and hybrid search with rank fusion sit behind one OpenAPI-described REST API, with API keys scoped to indices and privileges. The status page lists more than 20 incidents marked major since July 2026, and Elasticsearch 9.5.0 returned incorrect results for some queries without raising an error.\n- Choose it for: Retrieval that needs keyword relevance, filters and vectors in one query, and teams already running Elastic for logs or security.\n- Strength: OpenAPI 3.0.3 file for the Serverless API with 311 operations, plus llms.txt and a Markdown twin of every docs page\n- Strength: API keys take role descriptors that limit indices and privileges, with an expiry, and the MCP server accepts OAuth 2.1 on Serverless\n- Strength: Agent Builder MCP server is generally available on Serverless and on Elastic Stack 9.3 and later\n- Weakness: More than 20 status notices marked major between 10 July and 8 October 2026, three of them regional service incidents\n- Weakness: Elasticsearch 9.5.0 (29 July 2026) returned incorrect results for some `must_not` queries with no error, fixed in 9.5.1 and 9.5.2\n- Weakness: No request rate limit with numbers was found in the reviewed documentation\n- Price: $0.09 / unit · Auth: OAuth or key · x402: no · Where: local\n- Full assessment: https://www.anchorterminal.com/tools/elasticsearch.md\n- Against #1: https://www.anchorterminal.com/compare/elasticsearch-vs-meilisearch.md\n\n### 10. LanceDB, B 65.4/100\n\nOpen-source embedded database on the Lance columnar format.\n\n- Verdict: Embedded, no server, and tables can live directly in S3, GCS or Azure storage. No self-serve hosted API. Enterprise is sold through a contact form.\n- Choose it for: Agents whose code can import a library and keep retrieval data in object storage, from notebooks to multimodal datasets.\n- Strength: Embedded, no server, and tables can live directly in S3, GCS or Azure storage\n- Strength: Four releases since late July, v0.39.0 on 17 September 2026, with passing CI on main\n- Strength: Release notes with Breaking Changes and Deprecations sections, and minor-version bumps for breaking changes\n- Weakness: No self-serve hosted API. Enterprise is sold through a contact form\n- Weakness: No official MCP server, only a 3-tool reference example\n- Weakness: The docs' OpenAPI link for the Enterprise REST API returns 404\n- Price: Freemium · Auth: None · x402: no · Where: local\n- Full assessment: https://www.anchorterminal.com/tools/lancedb.md\n- Against #1: https://www.anchorterminal.com/compare/lancedb-vs-meilisearch.md\n\n4 more are ranked in the full table: https://www.anchorterminal.com/categories/vector-search.md\n\n## Head to head\n\n- [Meilisearch vs Pinecone API + MCP](https://www.anchorterminal.com/compare/meilisearch-vs-pinecone.md)\n- [Meilisearch vs Qdrant API + MCP](https://www.anchorterminal.com/compare/meilisearch-vs-qdrant.md)\n- [Meilisearch vs Typesense API + MCP](https://www.anchorterminal.com/compare/meilisearch-vs-typesense.md)\n- [Pinecone API + MCP vs Qdrant API + MCP](https://www.anchorterminal.com/compare/pinecone-vs-qdrant.md)\n- [Pinecone API + MCP vs Typesense API + MCP](https://www.anchorterminal.com/compare/pinecone-vs-typesense.md)\n- [Qdrant API + MCP vs Typesense API + MCP](https://www.anchorterminal.com/compare/qdrant-vs-typesense.md)\n\n## Questions\n\n### What are the highest-rated vector search and retrieval databases for AI agents?\n\nMeilisearch has the highest benchmark score of the 14 ranked vector search and retrieval databases, 78.6 (A). Pinecone API + MCP is second with 76.2 (BB).\n\n### How many vector search and retrieval databases are agent-ready?\n\n6 of the 14 ranked here grade BB or better, the bar for agent-ready on the Anchor benchmark.\n\n### Which vector search and retrieval databases accept x402 payments?\n\nNone of the ranked listings here accepts x402 for its main call yet.\n\n### Which of these vector search and retrieval databases is cheapest?\n\nBy published paid prices, Milvus and Zilliz Cloud API + MCP, at $0.025 per GB, the lowest of the 10 listings here with a paid price in this unit (free allowances aside). Plans, volume tiers and free allowances change the sum, so check the listing's price table.\n\n### How is this list ranked?\n\nBy the Anchor benchmark score out of 100, a weighted mean of the scored categories minus deductions for negative events, from public evidence re-checked as vendors change. Listings cannot pay for a place. The latest assessment behind this page is from 9 October 2026.\n\n## How this list is made\n\nThe order is the Anchor benchmark score, the same number as on each listing. Each listing is graded from public evidence against the benchmark checklist, and the picks are worked out from those grades, prices and facts. No listing pays for its place, and paid audits or listing help never change a score.\n", "meta": { "attribution": "Anchor Terminal (https://www.anchorterminal.com)", "docs": "https://www.anchorterminal.com/docs/", "generatedAt": "2026-10-10", "license": "CC-BY-4.0", "method": "https://www.anchorterminal.com/benchmark/", "methodology": "0.4", "openapi": "https://www.anchorterminal.com/openapi.json", "preview": false, "run": "2026-10-01", "runLabel": "October 2026 research run" }, "page": { "breadcrumbs": [ { "name": "Home", "url": "https://www.anchorterminal.com/" }, { "name": "Best of", "url": "https://www.anchorterminal.com/best/" }, { "name": "Retrieval \u0026 vector search", "url": "" } ], "description": "Meilisearch (A), Pinecone API + MCP (BB) and Supabase API + MCP (BB) lead the 14 ranked vector search and retrieval databases. Picks by need, strengths, weaknesses and prices from the Anchor benchmark.", "facts": [ "Meilisearch A", "Pinecone API + MCP BB", "Supabase API + MCP BB" ], "h1": "Best vector search and retrieval databases for AI agents", "image": "https://www.anchorterminal.com/assets/og/best-vector-search.png", "path": "/best/vector-search/", "published": "", "section": "tools", "title": "Best vector search and retrieval databases for AI agents in 2026", "toc": null, "updated": "2026-10-09", "url": "https://www.anchorterminal.com/best/vector-search/" }, "tokens": { "markdown": 5800, "slim": 1580 }, "version": 1 }