Chroma
trychroma.com
- Free plan
- Open source
- Windows
- macOS
- Linux
- Web
- iPhone and iPad
- Android
- API
- Self-hosted
What Chroma is
Chroma is open-source data infrastructure for AI that stores embeddings and metadata and supports search and retrieval across text, images, and other modalities. Its search options include dense, sparse, and hybrid vector search, full-text and regex search, and metadata filtering. It can run locally through pip, npm, or Docker, on Chroma Cloud, or in a customer’s own VPC. Documented embedding integrations include OpenAI, Cohere, Google Gemini, Hugging Face, Jina AI, and Ollama; framework integrations include LangChain, LlamaIndex, Haystack, Streamlit, and DeepEval. Chroma is licensed under Apache 2.0, and its maker says the same codebase powers the open-source database and Chroma Cloud. The cloud service lists encryption in transit and at rest, SSO, role-based access control, and employee MFA. Starter is free and includes 10 databases and 10 team members. Team costs $250 a month plus usage, includes $100 credits, and supports 100 databases and 30 team members. The pricing FAQ says service pauses when a customer-set or Chroma-set usage limit is exceeded, until the limit changes.
Who it suits
Chroma suits developers building AI search and retrieval systems who want local, cloud, or customer-VPC deployment. Teams evaluating its hosted plans should account for usage-based billing and the stated service pause when usage limits are exceeded.
Good to know, and what to watch
- Supports dense, sparse, and hybrid search.
- Offers local, cloud, and customer-VPC deployment.
- Licensed under Apache 2.0.
- Lists integrations for embedding services and frameworks.
- Cloud security includes encryption, SSO, and role-based access.
- Team costs $250 a month plus usage.
- Exceeding usage limits pauses service until limits change.
- Starter has 10 databases and 10 team members.
Plans
Every plan and priceTeam
Enterprise
Starter
Free plan
Getting started
- Open the Chroma website
- Choose local installation with pip, npm, or Docker, or select Chroma Cloud
- Connect an embedding provider such as OpenAI, Cohere, or Google Gemini
- Integrate a supported framework such as LangChain or LlamaIndex
- Select Starter, Team, or Enterprise deployment options
Limits to check
Starter includes 10 databases and 10 team members, with free credits followed by usage-based pricing. Team includes 100 databases, 30 team members, and $100 credits before usage charges. Customer-set or Chroma-set usage limits can pause service until changed; cloud specifications list 1 million collections per database and 5 million records per collection.
The Club's review
Overview
Chroma is open-source data infrastructure for storing embeddings and metadata and retrieving information across text, images, and other modalities. It is best suited to developers and teams building AI search who want both varied retrieval methods and a choice of local, hosted, or customer-controlled deployment. Its strongest case is flexibility; hosted teams need to account for usage-based charges and service pauses when limits are exceeded.
Key features
- Broad retrieval options: Dense, sparse, and hybrid vector search are joined by full-text, regex, and metadata filtering. This supports varied retrieval approaches in one system, though teams that need only a narrow vector-search setup may not benefit from the breadth.
- Index and dimension support: Chroma offers vector and sparse-vector indexes, full-text indexing, and inverted indexes for strings, integers, floats, and booleans. Its maximum vector size is 4096 dimensions.
- Integration choices: Documented embedding integrations include OpenAI, Cohere, Google Gemini, Hugging Face, Jina AI, and Ollama. Framework integrations include LangChain, LlamaIndex, Haystack, Streamlit, and DeepEval; SDKs are available for Python, TypeScript, Rust, Kotlin, and Swift. That range should help teams fit Chroma into varied AI stacks.
- Deployment and licensing: Apache 2.0 licensing and local options through pip, npm, or Docker make the open-source database accessible for development and self-hosting. Chroma Cloud and deployment in a customer’s own VPC add managed and customer-controlled paths; the maker says one codebase powers both the open-source database and Cloud.
- Cloud safeguards and scale: Chroma Cloud says it encrypts data in transit and at rest, supports SSO, role-based access control, and employee MFA, and completed a SOC 2 Type II examination. Enterprise deployment options include multi-tenant cloud, single-tenant nodes and clusters, BYOC, and on-premises deployments. Enterprise options also include single-region, multi-region, and multi-cloud regions, active-active and active-passive replication, and AWS, GCP, and Azure. Cloud technical specifications allow 1 million collections per database and 5 million records per collection.
Pricing
Chroma uses a freemium model. Starter is free, with 10 databases and 10 team members; it provides free credits before switching to usage-based pricing. That makes it a sensible starting point, but teams should plan for charges once credits run out.
Team costs $250 a month, plus usage, and includes $100 credits, 100 databases, 30 team members, Slack support, SOC II, and volume-based discounts. The fixed monthly charge buys substantially more database and seat capacity than Starter, but the additional usage cost means the monthly price is not the full bill. The pricing FAQ says exceeding customer-set or Chroma-set usage limits pauses service until the limit is changed, so teams must monitor limits to avoid interruption.
Enterprise has custom pricing and includes unlimited databases and team members. It is the fit for organizations that need those limits removed and enterprise deployment options; dedicated support is listed for this tier. Starter includes Community Slack, while Team includes Slack support.
Platforms
Chroma supports Android, iOS, Linux, macOS, Windows, web, API, and self-hosted use. Its listed SDK languages are Python, TypeScript, Rust, Kotlin, and Swift.
Who it's for
Chroma suits developers building retrieval for AI applications who want choices across vector, text, regex, and metadata search, plus the ability to move between local development, cloud hosting, and customer-controlled infrastructure. It is less compelling for teams seeking a hosted service with predictable flat pricing: Team adds usage charges to its monthly fee, and exceeded limits pause service.
Pros and cons
- Pro: Dense, sparse, hybrid, full-text, and regex search, along with metadata filtering, give teams several retrieval methods in one product.
- Pro: Apache 2.0 licensing and local, cloud, VPC, and on-premises deployment paths accommodate different infrastructure preferences.
- Pro: Integration options span several embedding providers, frameworks, and SDK languages.
- Con: Team’s $250 a month price is only a base charge; usage is additional.
- Con: Exceeding usage limits pauses the service until a limit is changed, making active quota management important for hosted workloads.
- Con: Starter is limited to 10 databases and 10 team members, while Team caps these at 100 databases and 30 team members.
Alternatives
For a different vector-database option, see Vector Databases; readers comparing broader database categories can also browse Embedded Databases, Search Databases, and Database Software.
- Vespa is a paid alternative with a free trial and metered Startup rates for vCPU, memory, disk, and GPU memory; consider it if those resource-based charges suit your cost model.
- Pinecone has a free plan with limits of five indexes, 100 namespaces per index, and up to 2 GB storage; it may suit teams prioritizing those stated free-tier allowances.
- Upstash Vector offers a free plan with 10K daily queries and updates, 200M vectors × dimensions, 1,536 maximum dimensions, 100 namespaces, and 1 GB of data and metadata; consider it when those quotas fit your workload.
- Zilliz Cloud has a free cluster with 5 GB storage, 2.5M vCUs per month, and up to five collections; it is an option if those free-cluster limits meet your needs.
- KDB.AI is another freemium option with cloud and self-hosted plans.
- Qdrant offers a free forever single-node cluster with 0.5 vCPU, 1 GB RAM, and 4 GB disk; consider it if a small, explicitly capped free cluster is sufficient.
- Weaviate has a free plan limited to one cluster, 100,000 objects, 1 GB memory, 10 GB disk, one collection, and up to three tenants; it may fit teams comfortable with those bounds.
- EmergentDB offers a freemium plan with a different approach to vector storage.
Verdict
Choose Chroma if you need flexible retrieval, open-source licensing, and a deployment path that can extend from local development to cloud or customer-controlled infrastructure. Its integrations and search range are strong reasons to consider it, but hosted teams should look elsewhere if a predictable flat bill or uninterrupted service beyond usage limits is essential.
The verdict
Chroma offers flexible search methods and deployment choices for AI data infrastructure, with an open-source license and a free Starter tier. Hosted teams should weigh the Team plan’s monthly charge plus usage and understand the service-pause rule for exceeded limits.
Questions people ask
Is Chroma open source?
Yes. Chroma is licensed under Apache 2.0, and the maker says the same codebase powers its open-source database and Chroma Cloud.
Where can Chroma run?
It can run locally through pip, npm, or Docker, on Chroma Cloud, or in a customer’s own VPC. Platforms listed include Android, iOS, Linux, macOS, Windows, web, API, and self-hosted use.
Which integrations are documented?
Embedding integrations include OpenAI, Cohere, Google Gemini, Hugging Face, Jina AI, and Ollama. Framework integrations include LangChain, LlamaIndex, Haystack, Streamlit, and DeepEval.
What support comes with each plan?
Starter lists Community Slack, Team lists Slack support, and Enterprise lists dedicated support.
What the maker publishes
- Changelog update
- The changelog lists EU region support in April 2026 and Chroma Sync for S3, GitHub and Web in March 2026. source
- CLI installation
- The CLI is installable through the Python package, JavaScript package, pipx or a standalone curl installer. source
- Cloud API
- The CloudClient accepts a tenant, database and API key, with environment-variable configuration also supported. source
- Cloud architecture
- Chroma Cloud is serverless and uses object storage with automatic tiering and caching. source
- Cloud performance limits
- The product page lists 1 million collections per database and 5 million records per collection as technical specifications. source
- Company founding
- Netfigo lists Chroma as founded in 2022 in San Francisco, with Jeff Huber and Anton Troynikov as founders. source
- Company legal name
- The terms identify the service provider as Chroma Inc. source
- Company location and size
- LinkedIn lists Chroma as privately held, headquartered in San Francisco and employing 11–50 people. source
- Compliance
- Chroma Cloud says it completed a SOC 2 Type II examination. source
- Dataset forking
- Forking supports dataset versioning, A/B tests and staged rollouts. source
- Deployment
- The maker says Chroma can run locally with pip, npm, or Docker, on Chroma Cloud, or in a customer’s own VPC. source
- Documentation depth
- The docs cover quickstarts, SDK references, cloud clients, deployment, Docker, production client-server mode, performance and migrations. source
- Embedding integrations
- Documented embedding integrations include OpenAI, Cohere, Google Gemini, Hugging Face, Jina AI, Ollama, and others. source
- Enterprise deployment
- Enterprise deployment options listed include multi-tenant cloud, single-tenant nodes and clusters, BYOC, and on-premises deployment. source
- Enterprise options
- The enterprise page lists single-region, multi-region, and multi-cloud regions, active-active and active-passive replication, and AWS, GCP, and Azure. source
- Framework integrations
- Documented framework integrations include LangChain, LlamaIndex, Haystack, Streamlit, and DeepEval. source
- Free plan
- Yes source
- Full-text and regex
- Its search interface includes trigram full-text matching and regular-expression queries. source
- Funding news
- Chroma announced an $18 million seed round led by Quiet Capital on April 7, 2023, and also disclosed a May 2022 pre-seed round. source
- Integrations: embedding providers
- Documented embedding integrations include OpenAI, Google Gemini, Cohere, Hugging Face, Jina AI, Ollama, Mistral and Together AI. source
- Integrations: frameworks
- Documented framework integrations include LangChain, LlamaIndex, DeepEval, Streamlit, Haystack, Google ADK, Mem0 and others. source
- Licence
- Chroma is released under the Apache 2.0 licence. source
- Limit: comparison performance
- A G2 reviewer says performance can fall below Pinecone in some scenarios. source
- Limit: documentation and learning curve
- A G2 reviewer cites limited advanced documentation, a potentially steep beginner learning curve and a focus narrower than general databases. source
Checked against the maker's pages on 15 Oct 2026.