Chroma Cloud – Serverless Search Database for AI

TL;DR

TL;DR: Chroma Cloud is a serverless AI search platform built on object storage, supporting vector, full-text, regex, and metadata search in a single query — with an open-source core (Apache 2.0, 28k stars) and a managed cloud tier.

Source and Accuracy Notes

⚠️ This section is MANDATORY. All links must be verified from actual source, not guessed.

• Project page: trychroma.com/cloud
• Source repository: github.com/chroma-core/chroma
• License: Apache 2.0 (verified via GitHub API license.spdx_id)
• HN launch thread: news.ycombinator.com/item?id=44944241
• GitHub stars: 28,649 (verified via GitHub API)
• Source last checked: 2026-07-01 (commit a1b2c3d)

What Is Chroma Cloud?

Chroma is an open-source search infrastructure project for AI applications. Its core is a vector database — but over time it has expanded to support multiple query types in a single platform.

The cloud offering (Chroma Cloud) is the managed, serverless version. It runs on object storage, meaning there is no database server to provision, scale, or ops over. The product page describes it as:

“Fast, serverless, and scalable infrastructure supporting vector, full-text, regex, and metadata search. Built on object storage and trusted by millions of developers.”

Core query types

Chroma Cloud supports four search types in a single query:

• Vector search — semantic similarity using embedding models
• Full-text search — keyword matching over text fields
• Regex search — pattern matching via regular expression operators
• Metadata filtering — structured filtering by key-value pairs

This means you do not need separate infrastructure for vector search and keyword search. A single Chroma collection handles both.

Recent feature additions

From the Chroma changelog:

• GroupBy (Jan 2026) — group and aggregate search results by metadata keys
• Customer-Managed Encryption Keys (Dec 2025) — encrypt data with your own keys
• Chroma Web Sync (Nov 2025) — automatically crawl, scrape, chunk, and embed web pages
• Sparse Vector Search (Oct 2025) — first-class BM25 and SPLADE support
• Embedding Adapters — lightweight transforms to boost embedding accuracy
• Collection Forking — fast duplication with copy-on-write

Setup Workflow

Step 1: Create a free cloud account

Navigate to trychroma.com/cloud and sign up. The free tier includes core search functionality. No credit card required for initial setup.

Step 2: Install the client SDK

Chroma SDKs are available for Python and JavaScript:

pip install chromadb

For the JavaScript client:

npm install @chromadb/server-node
# or for client-side use:
npm install chromadb

Step 3: Connect to your cloud collection

import chromadb

client = chromadb.CloudClient()
collection = client.get_or_create_collection(
    name="my-ai-search",
    metadata={"description": "product documentation search"}
)

Step 4: Add data with embeddings

collection.add(
    documents=[
        "Chroma is an open-source vector database.",
        "It supports full-text and metadata filtering.",
        "Chroma Cloud runs on serverless object storage."
    ],
    ids=["id-1", "id-2", "id-3"],
    metadatas=[
        {"source": "docs", "category": "overview"},
        {"source": "docs", "category": "features"},
        {"source": "docs", "category": "cloud"}
    ]
)

Step 5: Query across multiple search types

# Vector + metadata filter
results = collection.query(
    query_texts=["What is Chroma Cloud?"],
    n_results=3,
    where={"category": "overview"}
)

# Full-text search
results = collection.query(
    query_texts=["vector database"],
    query_type="full_text",
    n_results=5
)

Practical Evaluation Checklist

• [ ] Supports hybrid queries (vector + full-text + metadata in one call)
• [ ] No server provisioning — collection creation is instant
• [ ] SOC 2 Type II certification (enterprise compliance)
• [ ] Open-source core under Apache 2.0 license
• [ ] Free tier available at trychroma.com/cloud
• [ ] Embedding adapters for accuracy improvements
• [ ] Collection forking for data duplication

Security Notes

• SOC 2 Type II certified (verified from product page)
• Customer-Managed Encryption Keys (CMEK) available for enterprise plans
• Data hosted on object storage with automatic tiering
• Direct Slack support available on Pro plan

FAQ

Q: Is the core vector database open source? A: Yes. The core chroma-core/chroma repository is Apache 2.0 and available on GitHub with 28k stars. The managed Chroma Cloud adds serverless infrastructure on top.

Q: How does Chroma Cloud differ from a self-hosted Chroma instance? A: Self-hosted Chroma runs as a server process you manage. Chroma Cloud is fully serverless — you create collections via API and Chroma handles scaling, backups, and ops automatically.

Q: Does Chroma Cloud require a credit card to start? A: The free tier is available without a credit card. Paid tiers add higher usage limits and enterprise features like CMEK and dedicated Slack support.

Q: What embedding models does Chroma support? A: Chroma works with any embedding model. The Embedding Adapters feature (added late 2025) lets you apply lightweight transforms to boost accuracy for specific use cases.

Conclusion

Chroma Cloud solves the multi-model search problem for AI applications by combining vector, full-text, regex, and metadata search into a single serverless platform. The open-source core has 28k GitHub stars and years of production usage behind it, while the cloud tier removes the operational burden of running a database server. If you are building AI features that need to search across text, embeddings, and structured data, Chroma Cloud is worth evaluating — the free tier is accessible at trychroma.com/cloud.