Chroma
Open-source AI-native search infrastructure with a client-server architecture and pluggable storage backends. In embedded mode runs as SQLite + HNSW (via hnswlib); in server mode runs as a standalone gRPC/REST service. Cloud and self-hosted deployments use a tiered storage architecture — hot data in memory cache, warm data on SSD, cold data in S3/GCS object storage — with automatic query-aware data tiering managed by the runtime. Famous in the LangChain/LlamaIndex ecosystem as the lowest-friction vector database to spin up locally.
Definition
Open-source AI-native search infrastructure with a client-server architecture and pluggable storage backends. In embedded mode runs as SQLite + HNSW (via hnswlib); in server mode runs as a standalone gRPC/REST service. Cloud and self-hosted deployments use a tiered storage architecture — hot data in memory cache, warm data on SSD, cold data in S3/GCS object storage — with automatic query-aware data tiering managed by the runtime. Famous in the LangChain/LlamaIndex ecosystem as the lowest-friction vector database to spin up locally.
Most vector databases optimize for production at scale and treat developer ergonomics as secondary. Chroma inverted that: it's easy to embed in a Python notebook, has a tiny API surface, and works out-of-the-box for prototypes and small-to-medium production deployments. The 2026 framing has matured — Chroma is the right tool for single-node deployments up to ~5-10M vectors, but past that threshold the **Amnesia Loop** failure mode kicks in (retrieval timeouts → agent fallback to generic model knowledge → user-visible quality regression), and operators graduate to Qdrant, Milvus, or Pinecone.
Local prototyping and developer experimentation, RAG pipelines under 5-10M vectors, embedded vector search inside Python applications and notebooks, LangChain/LlamaIndex tutorial backends, and edge AI deployments where the entire vector store must fit on a single node.
Recent developments
Latest release: v1.5.9 (current as of July 2026). Tracking the upstream stable release line — note the
js_release_2.4.xtags in the repo are the JavaScript client's versioning, not the core engine. Per chroma-core/chroma releases.Chroma Cloud publishes pricing; Distributed Chroma goes BYOC. Chroma's managed tier now has public pricing, and Distributed Chroma is available bring-your-own-cloud — the "prototype ergonomics" product growing a managed production story aimed exactly at the 5-10M-vector graduation threshold described above. Per Chroma — pricing and Chroma — Distributed Chroma BYOC.
Tiered-storage architecture with S3/GCS cold tier. Chroma's Query Layer now uses memory cache (hot) + SSD cache (warm), backed by a Storage Layer that uses S3 / GCS (cold) for all vectors, metadata, and indexes — with automatic query-aware data tiering. Per trychroma.com products page.
Production scaling ceiling named explicitly. Independent comparisons now position Chroma as not production-grade at scale — performance degrades sharply past 10M vectors with the Amnesia Loop pattern as the failure signature. Operators are advised to graduate to Qdrant, Milvus, or Pinecone above that threshold. Per Encore guide and RankSquire 2026.
2026 migration playbook published. RankSquire published a 5-alternative migration ranking for teams outgrowing Chroma, with named landing spots for different access patterns. Per Chroma Database Alternative 2026.
DataRobot agentic-AI integration. DataRobot now ships Chroma as the default self-hosted vector DB for its agentic-AI runtime, with explicit S3-backed deployment docs. Per docs.datarobot.com.
v1.5.9 detail: sharding upgrades plus a security warning (tertiary). The May 2026 release adds
GROUP BYsupport for sharded collections and a sharded-collection rebuild capability; independent security research separately reported a pre-auth vulnerability affecting older self-hosted instances, with guidance to run v1.5.9 and keep instances off the public network. Per Vector Database News May 2026 (tertiary, not corroborated against an official Chroma security advisory).The reported "2.4.2" major-version jump is a separate false positive from a dev-build suffix —
1.5.10.dev242, not a release scrape artifact. The242integer traces to1.5.10.dev242, a pre-release development build tagged August 4, 2026, following the stable1.5.9release in May 2026 already noted above. Chroma's core architecture remains firmly in the 1.x line, with stable API compatibility maintained since the1.0.15release in July 2025. Per chroma-core/chroma releases.Client-side API now splits EphemeralClient (in-memory) from PersistentClient (directory-backed). The two paradigms target edge deployments explicitly: an ephemeral instance for throwaway/in-process use, a persistent one backed by a local directory when state must survive a restart — rather than one client class inferring intent from arguments. Per Chroma docs.
The v1 API was fully deprecated at v0.7.0 — v2 is now the only supported interface — and a dedicated MCP server exposes Chroma directly to LLM agents. The chroma-core/chroma-mcp server lets agents query/write collections over the Model Context Protocol without a client SDK in the loop. Per chroma-core/chroma releases. Sources: trychroma.com · GitHub (chroma-core/chroma) · Chroma docs · RankSquire — Best Self-Hosted Vector DB 2026 · DataRobot — ChromaDB VDB docs