Retrieval & Data Infrastructure

Search platforms (non-memory)

Hosted search API ↗ 2

Managed-only — hosted search-as-a-service API ↗ ●

Free + paid ↗ 2

namespace ↗ 2

searched not found ↗ 2

via official adapter — Algolia MCP ↗ 2

proprietary (hosted SaaS); open-source API clients/codegen are MIT ↗ ●

not applicable — managed service (JS API client algoliasearch v5.60.0) ↗ ●

2026-10-26

Knowledge-graph platforms

RDF triple/quad store + reasoning ↗ 2

Both (on-prem + cloud) ↗ 2

Free tier + commercial licenses (quote-based) ↗ 2

Repository-level access control (statement-level Security Filters per user/role); on-prem only for full isolation ↗ 2

not documented in public vendor materials (operational dimension typically managed in private deployment guides). ↗ 2

no first-party MCP adapter published as of 2026-05; community connectors may exist. ↗ 2

2026-10-27

Knowledge-graph platforms

Vector + graph in one managed service ↗ 2

Managed-only ↗ 2

Pay-per-use — billed per m-NCU-hour (1 m-NCU = 1 GB memory plus compute for 1 hour), $0.03 per m-NCU-hour in us-east-1 (16 m-NCU graph = $0.48/hour); paused graphs billed at 10% of compute ↗ 2

hard-isolation ↗ 2

not documented in public vendor materials (operational dimension typically managed in private deployment guides). ↗ 2

yes — official AWS Labs Amazon Neptune MCP server (openCypher queries, schema and status for Neptune Analytics graphs) ↗ ●

2026-10-27

Knowledge-graph platforms

Multi-model: graph + doc + KV + vector ↗ 2

Both ↗ 2

Pay-per-use ↗ 2

hard-isolation ↗ 2

not documented in public vendor materials (operational dimension typically managed in private deployment guides). ↗ 2

yes — community ravenwits/mcp-server-arangodb ↗ ●

BSL-1.1 (source-available; each release converts to Apache-2.0 four years after release) ↗ ●

2026-10-29

Vector-database infrastructure

Embedded vector DB (Rust core) ↗ 2

Both (self-hosted + managed cloud) — OSS embedded/server; Chroma Cloud serverless; single-tenant/BYOC on Enterprise ↗ 2

Chroma Cloud winding down: read-only from 2026-11-22, no reads after 2026-12-22, no new Cloud plans; open-source ChromaDB free (Apache-2.0) ↗ 2

Multi-tenant indexes with billions of vectors; AWS PrivateLink; BYOC (Bring Your Own Cloud) for hard isolation ↗ 2

strong ↗ 2

yes — official chroma-core/chroma-mcp (GitHub, Apache-2.0) ↗ ●

Apache-2.0 ↗ ●

v1.6.0 (2026-10-13) ↗ ●

2026-10-29

Enterprise-search adjacencies

Curated bibliographic metadata ↗ 2

Both ↗ 2

Enterprise only ↗ 2

searched not found ↗ 2

not documented in public vendor materials (operational dimension typically managed in private deployment guides). ↗ 2

yes — IPOne MCP servers: Derwent Patent MCP, CompuMark Trademark MCP, Darts-ip Litigation MCP ↗ 2

2026-10-27

Embedding & reranker services

Hosted embeddings API ↗ 2

Both (self-hosted + managed cloud) — Cohere API and Model Vault; deployable in customer cloud via Amazon SageMaker, Azure AI Foundry and Oracle OCI ↗ 2

Model Vault dedicated instances: Embed 5 Fast/Pro Small $3.00/hour ($2,000/month), Medium $5.00/hour ($3,250/month); Rerank 5 Fast Medium $4.00/hour ($2,500/month); embed models billed per token embedded, rerank per search on the API ↗ 2

searched not found ↗ 2

searched not found ↗ 2

proprietary; Python SDK MIT ↗ 2

Embed 5 (2026-09-30) ↗ 2

2026-10-28

Enterprise-search adjacencies

RAG-as-a-Service via hosted MCP ↗ 2

Managed-only ↗ 2

Enterprise only ↗ 2

Logical multi-tenancy on AWS — dedicated Coveo Index per customer for full data segmentation ↗ 2

not documented in public vendor materials (operational dimension typically managed in private deployment guides). ↗ 2

yes — hosted Coveo MCP Server connecting AI agents (ChatGPT, Claude, Copilot) to Coveo retrieval ↗ 2

2026-10-27

Vector-database infrastructure

Storage-optimised + Unity Catalog governance ↗ 2

Managed-only ↗ 2

Pay-per-use DBUs — Standard $0.28/hour per unit (2M 768-dim vectors, 4 DBU/hour); Storage Optimized $1.28/hour per unit (64M vectors, 18.29 DBU/hour), effective US East prices ↗ 2

Workspace-level isolation; separate CMEK per business unit/environment for encryption isolation; revoking key renders data inaccessible ↗ 2

not documented in public vendor materials (operational dimension typically managed in private deployment guides). ↗ 2

yes — Databricks managed MCP server for AI Search indexes (/api/2.0/mcp/ai-search/{catalog}/{schema}/{index}) ↗ 2

2026-10-29

Knowledge-graph platforms

Auto-extracted KG from web crawl ↗ 2

Managed-only ↗ 2

Free $0/month (10,000 credits); Startup $299/month (250,000 credits, overage $0.001/credit); Startup Plus $899/month (1,000,000 credits, overage $0.0009/credit); Enterprise custom ↗ 2

hard-isolation ↗ 2

not documented in public vendor materials (operational dimension typically managed in private deployment guides). ↗ 2

yes — official diffbot/diffbot-mcp server (extract, search_web, crawl, entity resolution tools) ↗ ●

2026-10-27

Vector-database infrastructure

Mature hybrid search ↗ 2

Both ↗ 2

free (OSS) self-managed; Elastic Cloud paid (price pages not re-verified) ↗ ●

Allocator-level cluster separation in Elastic Cloud Enterprise; index-level access control; dedicated single-tenant deployments available ↗ 2

not documented in public vendor materials (operational dimension typically managed in private deployment guides). ↗ 2

yes — official Elastic Agent Builder MCP endpoint (Elastic 9.2+ / Serverless); older elastic/mcp-server-elasticsearch deprecated ↗ ●

AGPL-3.0 / SSPL-1.0 / Elastic-2.0 (choice) ↗ ●

not applicable — managed service ↗ ●

2026-10-26

Enterprise-search adjacencies

Enterprise search + per-user KG ↗ 2

Both ↗ 2

Enterprise only ↗ 2

Logically isolated single-tenant per customer (data, models, telemetry siloed; no shared vector index); option for fully isolated single-tenant in customer AWS/Azure/GCP ↗ 2

not documented in public vendor materials (operational dimension typically managed in private deployment guides). ↗ 2

native (first-party) — Glean MCP server ↗ 2

2026-10-27

Embedding & reranker services

HF hosted embeddings ↗ 2

Both (self-hosted + managed cloud) — managed Inference Providers and Inference Endpoints; self-hosted Text Embeddings Inference (TEI) ↗ ●

Inference Providers pay-as-you-go at provider rates with no Hugging Face markup; PRO users get $2.00/month credits, Team/Enterprise $2.00 per seat; Inference Endpoints billed hourly per instance ↗ 2

searched not found ↗ 2

yes — official Hugging Face MCP Server (huggingface/hf-mcp-server) ↗ 2

proprietary service; huggingface_hub SDK Apache-2.0; Text Embeddings Inference Apache-2.0 ↗ ●

v2.3.0 (2026-10-23) — huggingface_hub ↗ 2

2026-10-27

Embedding & reranker services

Open embedding models + API ↗ 2

Both (self-hosted + managed cloud) — hosted Jina API; open weights and jina-on-prem air-gapped Docker/K8s; SageMaker ↗ ●

Hosted API: prepaid token packages with a free trial key and optional auto-recharge (pricing model since 2025-05-06); also on AWS SageMaker and Microsoft Azure; Jina On-Prem containers sold through Elastic Sales ↗ 2

searched not found ↗ 2

yes — official jina-ai/MCP remote server (mcp.jina.ai/v1; embeddings, rerank, reader, search tools) ↗ ●

mixed — most recent open-weight models CC-BY-NC-4.0 (v3, v5-omni, v5-text, clip-v2, colbert-v2, reranker-v3; commercial licence via Elastic); jina-embeddings-v4 under the Qwen Research License (its model card says the initial CC-BY-NC-4.0 label was an error); some older models Apache-2.0 ↗ 2

jina-embeddings-v5-omni (nano and small; 2026-05-07); reranker jina-reranker-v3.5 (2026-07-27) ↗ 2

2026-10-29

Vector-database infrastructure

Embedded Arrow-columnar vector DB ↗ 2

Both (self-hosted + managed cloud) — embedded OSS library; LanceDB Enterprise as Managed or BYOC ↗ 2

free (OSS); LanceDB Enterprise/Cloud pricing via sales (no public price list) ↗ 2

Logical namespace per dataset; embedded library or LanceDB Cloud (single-tenant available) ↗ 2

strong ↗ 2

yes — official reference server lancedb/lancedb-mcp-server (basic, 3 tools) ↗ ●

Apache-2.0 ↗ ●

v0.40.0 (2026-10-07) ↗ 2

2026-10-26

Enterprise-search adjacencies

Session-history conversational memory atop RAG ↗ 2

Both (self-hosted + managed cloud) — Lucidworks Platform SaaS; self-hosted Fusion ↗ 2

searched not found ↗ 2

Per-tenant index hard-isolation in Fusion Cloud ↗ 2

yes — Lucidworks MCP server exposes Fusion query pipelines to Claude and other MCP clients ↗ 2

2026-10-27

Vector-database infrastructure

Multi-vector + hybrid + GPU ↗ 2

Both (self-hosted OSS + Zilliz Cloud managed) ↗ 2

free (OSS); managed via Zilliz Cloud (free tier, then usage-based) ↗ 2

Control plane and data plane in separate VPCs/subnets; BYOC option for hard tenant isolation ↗ 2

eventual ↗ 2

yes — official zilliztech/mcp-server-milvus (GitHub) ↗ ●

Apache-2.0 ↗ ●

v3.0.3 (2026-10-14) ↗ 2

2026-10-26

Vector-database infrastructure

Vector embedded in document DB ↗ 2

Both (self-hosted + managed cloud) — Atlas managed; Atlas Local deployments with vector search (auto-embedding via Voyage AI) ↗ ●

Pay-per-use ↗ 2

Project/cluster/database isolation; dedicated cluster tier; PrivateLink; JWT-based tenant scoping for multi-tenant apps ↗ 2

eventual ↗ 2

yes — official mongodb-js/mongodb-mcp-server (Apache-2.0) ↗ ●

2026-10-26

Knowledge-graph platforms

Property graph + vectors + Aura Agent ↗ 2

Both (self-hosted + managed cloud) — Community/Enterprise self-managed; Neo4j Aura cloud ↗ ●

Free + paid ↗ 2

AuraDB Virtual Dedicated Cloud / AuraDS Enterprise: dedicated AWS/Azure/GCP account/subscription/project per customer; PrivateLink supported ↗ 2

strong ↗ 2

yes — official neo4j/mcp (product team); Neo4j Labs also ships neo4j-contrib/mcp-neo4j ↗ ●

GPL-3.0 (Community Edition); Enterprise Edition proprietary ↗ ●

2026.07 CalVer series (current default branch; exact patch and date not captured) ↗ ●

2026-10-26

Knowledge-graph platforms

RDF store + real-time inference ↗ 2

Both ↗ 2

Enterprise only ↗ 2

hard-isolation ↗ 2

not documented in public vendor materials (operational dimension typically managed in private deployment guides). ↗ 2

yes — GraphDB 11.4 supports custom MCP prompt templates (defined via REST) for AI agents ↗ 2

proprietary; GraphDB Free (free-to-use non-commercial option) and Enterprise Edition; every edition needs a licence file (Free licence emailed on request) ↗ 2

2026-10-29

Embedding & reranker services

OpenAI embedding API ↗ 2

Managed-only — OpenAI API ↗ 2

text-embedding-3-small $0.02 per 1M tokens; text-embedding-3-large $0.13 per 1M tokens ↗ 2

searched not found ↗ 2

searched not found ↗ 2

proprietary; SDKs Apache-2.0 ↗ ●

v3.39.0 (2026-10-26) — openai (PyPI) ↗ 2

2026-10-27

Search platforms (non-memory)

Open Elasticsearch fork ↗ 2

searched not found ↗ 2

free (OSS); Amazon OpenSearch Service billed separately by AWS ↗ ●

searched not found ↗ 2

searched not found ↗ 2

yes — official opensearch-project/opensearch-mcp-server-py (Apache-2.0) ↗ ●

Apache-2.0 ↗ 2

searched not found ↗ 2

2026-10-26

Knowledge-graph platforms

Operational digital twin ↗ 2

Both ↗ 2

Enterprise only ↗ 2

Foundry tenant per customer with strong ACL classification; zero-trust architecture; dedicated FedRAMP High enclave for federal workloads ↗ 2

not documented in public vendor materials (operational dimension typically managed in private deployment guides). ↗ 2

yes — Ontology MCP (exposes Ontology to external agents) and Palantir MCP (palantir/palantir-mcp) ↗ 2

2026-10-29

Vector-database infrastructure

PostgreSQL vector extension ↗ ●

Self-hosted only — PostgreSQL extension; also offered inside managed Postgres (Supabase, AWS RDS, Azure, Neon, etc.) ↗ ●

free (OSS) ↗ ●

Logical isolation via Postgres schemas/databases; managed providers (Supabase, Neon, RDS) add tenant-level controls ↗ ●

strong ↗ 2

yes — community Postgres MCP servers with pgvector tools (e.g. MOmayr/mcp-postgres-pgvector) ↗ ●

PostgreSQL ↗ ●

v0.8.6 (2026-07-29) ↗ ●

2026-10-26

Vector-database infrastructure

Managed vector DB + cascading retrieval ↗ 2

Managed-only — serverless SaaS on AWS/GCP/Azure; BYOC (GA 2026-09) runs in the customer's cloud account ↗ 2

Starter free (2 GB, 2M WU/1M RU per mo); Builder $20/mo; Standard $50/mo minimum; Enterprise $500/mo minimum ↗ 2

Logical resource isolation in serverless; CMEK additionally per-namespace key separation; project + organization RBAC ↗ 2

eventual ↗ 2

yes — official @pinecone-database/mcp (npm) plus per-Assistant remote MCP servers ↗ 2

2026-10-26

Vector-database infrastructure

Vector DB + DBSF + RRF ↗ 2

Both (self-hosted + managed cloud) — OSS engine; Qdrant Managed Cloud, Hybrid Cloud and Private Cloud ↗ 2

Free tier forever (single node, 1 GB RAM / 4 GB disk); Standard usage-based hourly billing; Premium has minimum spend; Hybrid/Private Cloud via sales ↗ 2

Hardened unprivileged container per cluster with strict network policies; outbound restricted; paid plans on dedicated resources; Private Cloud option for air-gap ↗ 2

eventual ↗ 2

yes — official qdrant/mcp-server-qdrant (GitHub, Apache-2.0) ↗ ●

Apache-2.0 ↗ ●

v1.19.2 (2026-10-05) ↗ ●

2026-10-26

Enterprise-search adjacencies

Agentic enterprise search + KCS ↗ 2

Managed-only ↗ 2

Enterprise only ↗ 2

hard-isolation ↗ 2

not documented in public vendor materials (operational dimension typically managed in private deployment guides). ↗ 2

yes — official SearchUnify MCP server (su-mcp) exposing search and analytics tools ↗ 2

2026-10-27

Embedding & reranker services

Open embedding training framework ↗ 2

Self-hosted only — Python library (pip) ↗ ●

free (OSS) ↗ ●

searched not found ↗ 2

searched not found ↗ 2

Apache-2.0 ↗ 2

searched not found ↗ 2

2026-10-26

Enterprise-search adjacencies

Hybrid retrieval + agentic RAG ↗ 2

Both ↗ 2

Enterprise only ↗ 2

On-premises / private-cloud tenant / fully-managed SaaS — customer choice; Azure-region selectable ↗ 2

not documented in public vendor materials (operational dimension typically managed in private deployment guides). ↗ 2

yes — Sinequa MCP Server exposing 200+ connectors and search to MCP clients ↗ 2

2026-10-28

Vector-database infrastructure

Hybrid vector + keyword + semantic rerank ↗ 2

Managed-only ↗ 2

Pay-per-use ↗ 2

Indexing runs single-tenant in customer Snowflake account; multi-tenant Cortex Agents enforce isolation via immutable session attributes + row access policies ↗ 2

not documented in public vendor materials (operational dimension typically managed in private deployment guides). ↗ 2

yes — official Snowflake-managed MCP server (Cortex Agents MCP); Snowflake-Labs/mcp community server deprecated ↗ ●

2026-10-26

Knowledge-graph platforms

RDF semantic platform + Voicebox AI ↗ 2

Both ↗ 2

Stardog Free (renewable 1-year license, commercial use allowed, no HA/caching/backups); Enterprise on-prem and cloud quote-based ↗ 2

Per-database access control; on-prem / private-cloud / SaaS options ↗ 2

not documented in public vendor materials (operational dimension typically managed in private deployment guides). ↗ 2

yes — official Stardog Cloud MCP server (stardog-union/stardog-cloud-mcp; remote setup in beta) ↗ ●

2026-10-27

Search platforms (non-memory)

Postgres + pgvector cloud ↗ 2

Both (self-hosted + managed cloud) — Supabase Cloud or self-hosted Supabase stack ↗ ●

searched not found ↗ 2

searched not found ↗ 2

searched not found ↗ 2

yes — official supabase/mcp (Apache-2.0) ↗ ●

Apache-2.0 (Supabase stack); vectors via pgvector (PostgreSQL licence) ↗ ●

searched not found ↗ 2

2026-10-29