Chroma Documentation
by Chroma
Open-source retrieval infrastructure, with docs that actually teach retrieval
Overview
The docs are split into Overview, Collections, Querying Collections, Embeddings, Run Chroma, Cloud, CLI, Guides, Integrations and Reference. Collections covers creating, configuring, updating and deleting data plus conditional transactions; Querying covers query and get, full-text search and metadata filtering. The Cloud section documents the serverless offering in depth — Search API basics, batch operations, filtering, group-by and aggregation, hybrid search with reciprocal rank fusion, ranking and scoring, pagination and field selection, collection forking, schema and index configuration, sparse-vector setup, quotas and sync from GitHub, S3, file upload or the web. The Guides section is what separates this from typical vendor docs: Build includes Intro to Retrieval, Building with AI, Agentic Search, Agentic Memory, Chunking and Look at Your Data, so it teaches retrieval design rather than only the API surface; Deploy covers Docker, client-server mode, AWS, Azure, GCP, observability and thin-client; Performance splits single-node from distributed and Cloud. Reference carries language bindings for Python, TypeScript, Rust, Kotlin and Swift plus architecture notes for the distributed build. Integrations list 25+ embedding providers and 12+ frameworks. The project is Apache-2.0 at 28.9k GitHub stars with a Rust core, ships package releases weekly on Mondays, and Chroma Cloud is generally available with $5 of free credits and usage-based pricing after that.
At a Glance
- Topic
- RAG
- Level
- All Levels
- Format
- Documentation
- Cost
- Freemium
- Duration
- Self-paced reference; ~1-2 hours from the Getting Started guide to a working retrieval pipeline
- Provider
- Chroma
- Hands-on
- Yes — code/exercises
- Certificate
- None
What You’ll Learn
- ✓Create, configure and fork collections and manage them from the Chroma CLI
- ✓Run dense, sparse and hybrid vector search with reciprocal rank fusion ranking
- ✓Combine full-text search, regex search and metadata where-filters in a single query
- ✓Choose and swap embedding functions across 25+ providers, including multimodal embeddings
- ✓Chunk documents sensibly and inspect retrieval quality using the Look at Your Data guide
- ✓Deploy Chroma with Docker or on AWS, Azure and GCP in client-server or thin-client mode
- ✓Build agentic search and agentic memory on top of the Chroma Search API
- ✓Tune single-node versus distributed performance and sync collections from S3, GitHub or the web
Highlights
- •28.9k GitHub stars, Apache-2.0, with a Rust core powering the distributed and Cloud deployments
- •The Guides section teaches retrieval design — Intro to Retrieval, Chunking, Look at Your Data — not just API calls
- •First-party language references for Python, TypeScript, Rust, Kotlin and Swift, so it is not Python-only
- •The same docs cover local, self-hosted and serverless Cloud paths, so a prototype does not need rewriting to ship
- •Weekly Monday release cadence keeps the documentation close to the shipped code
Who It’s For
Best For
- ✓Engineers building a first RAG pipeline who want to start locally with pip install
- ✓Teams evaluating vector databases against Qdrant, Weaviate, Milvus or pgvector
- ✓Developers adding retrieval memory to agents rather than to a chatbot
- ✓Anyone needing hybrid dense plus sparse plus full-text search in one store
Prerequisites
- •Working Python or TypeScript and comfort with package installation
- •A basic grasp of embeddings and vector similarity search
- •Access to an embedding provider API key, or the ability to run sentence-transformers locally
FAQ
What is Chroma Documentation?
The official documentation for Chroma, the Apache-2.0 open-source search infrastructure behind a large share of RAG prototypes. It targets engineers building retrieval into LLM applications and agents, covering dense, sparse and hybrid vector search, full-text and regex search, metadata filtering and multimodal indexing. After working through it you can stand up a local collection, tune chunking and embeddings, and deploy the same code to Docker, a cloud VM or Chroma Cloud.
Is Chroma Documentation free?
Chroma Documentation offers free content, with paid options for certificates or premium features.
What level is Chroma Documentation for?
Chroma Documentation is aimed at a all levels audience. Recommended background: Working Python or TypeScript and comfort with package installation, A basic grasp of embeddings and vector similarity search, Access to an embedding provider API key, or the ability to run sentence-transformers locally.
How long does Chroma Documentation take?
Expect roughly Self-paced reference; ~1-2 hours from the Getting Started guide to a working retrieval pipeline. Most learners work through it at their own pace.
What will I learn from Chroma Documentation?
You'll learn: Create, configure and fork collections and manage them from the Chroma CLI; Run dense, sparse and hybrid vector search with reciprocal rank fusion ranking; Combine full-text search, regex search and metadata where-filters in a single query; Choose and swap embedding functions across 25+ providers, including multimodal embeddings; Chunk documents sensibly and inspect retrieval quality using the Look at Your Data guide; Deploy Chroma with Docker or on AWS, Azure and GCP in client-server or thin-client mode; Build agentic search and agentic memory on top of the Chroma Search API; Tune single-node versus distributed performance and sync collections from S3, GitHub or the web.
Topics
Sources
This page was written from 3 sources, 1 on domains other than docs.trychroma.com.