Open-source data infrastructure for AI
Fast, serverless, and scalable infrastructure supporting vector, full-text, regex, and metadata search. Built on object storage and trusted by millions of developers. Open-source Apache 2.0.
Or, get started locally.
AI App
Chroma
knowledge_base - 1,277,467 records
awaiting query input
15M+ monthly downloads
Apache 2.0
27k Github stars
27k Github stars
Low latency search
Fast queries over billions of multi-tenant indexes.
Up to 10x cheaper
Built on object storage with automatic data tiering.
No engineering ops
Scales with your data and traffic. SOC 2 Type II.
Features
◆
Vector search
Semantic similarity search
◇
Sparse vector search
Lexical search (BM25, SPLADE)
●
Full-text search
Trigram and regex search
◐
Metadata search
Filtering and faceted search
◊
Forking
Dataset versioning, A/B testing, and roll-outs
▣
CLI
Command-line tools for development
import { ChromaClient } from 'chromadb'
const client = new ChromaClient()
const collection = await client.getOrCreateCollection({
name: "my_collection"
})
// Add documents with embeddings
await collection.add({
ids: ["id1", "id2"],
documents: ["This is a document", "Another doc"],
embeddings: [[1.2, 2.3, ...], [3.4, 4.5, ...]]
})
// Query by vector similarity
const results = await collection.query({
queryEmbeddings: [[1.1, 2.2, ...]],
nResults: 10
})Terminal Output
Run the code above to see the output ^
Performance
Fast search over billions of multi-tenant indexes
Chroma's indexes are built and optimized for object-storage offering unparalleled cost and performance. State-of-the-art vector, full-text, and regex search.
Latency
Query Latency
@384 dim at 100k vectors
Warm
Cold
p50
20ms
650ms
p90
27ms
1.2s
p99
57ms
1.5s
Contact us to run a POC for your specific workload.
Dedicated clusters can be scaled to your specific requirements.
Dedicated clusters can be scaled to your specific requirements.
Technical specs
Write throughput (per collection)30 MB/s (2000+ QPS)
Concurrent reads (per collection)10 (200+ QPS)
Collections per database1M
Records per collection5M
Recall90-100%
Zero-ops infra
┌───────────────────────────────┐ │ Query Layer │ │ Fast memory cache (hot) │ │ SSD cache (warm) │ └───────────────────────────────┘ ↕ Intelligent tiering ┌───────────────────────────────┐ │ Storage Layer │ │ S3 / GCS (cold) │ │ • All vectors │ │ • All metadata │ │ • All indexes │ └───────────────────────────────┘
Unlike legacy search systems, Chroma is a database you'll want to be on-call for.
✓Auto-scales with usage
✓No manual tuning
✓Serverless pricing
Chroma takes full advantage of object storage with automatic query-aware data tiering and caching.
✓Vectors are large: 1GB text → 15GB of vectors
✓Memory is expensive: $5/GB/mo
✓Object storage is not: $0.02/GB/mo
Enterprise
Chroma brings the security, compliance, education and operational model enterprises need with our Apache 2.0 architecture.
BYOC in your VPC, multi-cloud/multi-region replication, point-in-time-recovery ensure a resilient and scalable search system with the same 0-ops story as Cloud.
▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓ ▓░ ░▓ ▓░ ┌──────────── YOUR VPC ─────────────┐ ░▓ ▓░ │ │ ░▓ ▓░ │ █ DATA PLANE █ │ ░▓ ▓░ │ │ ░▓ ▓░ │ Your data, your cloud │ ░▓ ▓░ │ │ ░▓ ▓░ │ │ ░▓ ▓░ └───────────────────────────────────┘ ░▓ ▓░ │ ░▓ ▓░ │ ░▓ ▓░ ▼ ░▓ ▓░ ═════════════════════════════════════ ░▓ ▓░ ░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░ ░▓ ▓░ ░▓ ▓░ ┌────────── CHROMA VPC ─────────────┐ ░▓ ▓░ │ │ ░▓ ▓░ │ █ CONTROL PLANE █ │ ░▓ ▓░ │ │ ░▓ ▓░ │ Managed by Chroma │ ░▓ ▓░ │ Monitoring, backups, ops │ ░▓ ▓░ │ │ ░▓ ▓░ └───────────────────────────────────┘ ░▓ ▓░ ░▓ ▓░ ✓ BYOC in your VPC ░▓ ▓░ ✓ Multi-region replication ░▓ ▓░ ✓ 0-ops management ░▓ ▓░ ░▓ ▓░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░▓ ▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒
[▶] Videos
Chroma Context-1
13:20
Lexical Search in Chroma
4:41
Schema() and Search() APIs
9:02
Context Engineering Episode 3 - Lance Martin - LangChain
1:02:36
Beyond The Embedding: Vector Indexing
11:26
Long live Context Engineering
57:00
Context Rot
7:55
Context Engineering: The Outer Loop
23:43
Context Engineering for Engineers
11:16
Reliability at Scale
26:30
Context Engineering with DSPy
12:46
See more
Visit our YouTube channel →
[●] Open source community
Open-source databases give your team the control and flexibility to build exactly what you need. No licensing limits, no vendor lock-in, just reliable performance backed by a large community.
Github →
Chroma has over 26k GitHub stars and is used in over 90k other open-source codebases on GitHub. It is downloaded over 11M times a month.
Discord →
Join the Discord to see what people are building!
Run Chroma OSS →
Run Chroma on your own infrastructure with our open-source deployment guides.
[◆] Support
Open-source →
Join our 10K person strong Discord community to get fast and expert help from the open-source community.
All plans →
Helpful support direct from engineers on the Chroma team
Pro plan →
Direct Slack communication for fast support and help designing and iterating your search system.
Enterprise plan →
Customized SLAs ensure your team gets 24/7 assistance.
[▲] Research
Our research spans both basic and applied research for search, retrieval, agents, and context engineering.
Context-1
Training a self-editing search agent.
Context Rot
How increasing input tokens impacts LLM performance.
Generative Benchmarking
New methods for evaluating retrieval systems.
Chunking Strategies
Evaluating chunking strategies in retrieval for AI.
Embedding Adapters
Lightweight transforms to boost embedding accuracy.
[■] Updates
Chroma's project is rapidly improving. Here are the latest updates.
We’re looking for curious people who are dedicated to becoming world-class at their craft to join our team.
Get started
Get up and running in 30 seconds or less with $5 in free credits.
Quick Start
pip install chromadbJavaScript / TypeScriptJavaScript / TypeScript getting started docs →
npm install chromadb
