Skip to content

Best Free Vector Databases of 2026

Updated · 5 picks · live pricing · affiliate disclosure

Embedded vector DB with Lance columnar format and S3-compatible storage backend.

BEST OVERALL7.5/10

LanceDB

Embedded vector DB with Lance columnar format and S3-compatible storage backend.

OSS Apache 2.0 free; Cloud trial available

How it stacks up

  • OSS embedded free

    vs Chroma single-node

  • Cloud trial credits

    vs Qdrant Rust OSS

  • Pay-per-query

    vs Weaviate hybrid

#2
Weaviate5.3/10

From $25/mo

View
#3
Qdrant4.8/10

From $25/mo

View

All picks at a glance

#PickBest forStartingScore
1LanceDBBest free embedded analytics, columnar Lance with S3 storageFree7.5/10
2WeaviateBest free hybrid search OSS, vector plus keyword with GraphQL$25.00/mo5.3/10
3QdrantBest free Rust-based production, filter-aware HNSW indexing$25.00/mo4.8/10
4pgvector (Postgres)Best free Postgres extension, vector search inside existing Postgres$19.00/mo4.6/10
5ChromaBest free for prototyping, Python-first single-node embedded$25.00/mo4.2/10

Quick pick by use case

If you only have thirty seconds, find your situation below and skip to that pick.

Compare all 5 picks

Top spec
#1LanceDB7.5/10FreeOSS embedded free
#2Weaviate5.3/10$25.00/mo$300.00/yr$240/yr moreOSS BSD-3 free
#3Qdrant4.8/10$25.00/mo$300.00/yr$240/yr moreOSS Apache 2.0 free
#4pgvector (Postgres)4.6/10$19.00/mo$228.00/yr$168/yr moreOSS extension free
#5Chroma4.2/10$25.00/mo$300.00/yr$240/yr moreOSS embedded free
#1

LanceDB

7.5/10

Best free embedded analytics, columnar Lance with S3 storage

Embedded vector DB with Lance columnar format and S3-compatible storage backend.

PlanMonthlyWhat you get
OSS embeddedFreeApache 2.0 embedded serverless DB with Lance columnar format and Arrow.
LanceDB CloudFreeFree trial of hosted managed LanceDB with S3-compatible storage.
EnterpriseCustomCustom pricing with BYOC option, custom SLA, and dedicated support.

LanceDB is the embedded analytics free pick and the right call for analytics teams that want serverless vector storage with S3 backend. Founded 2022 in San Francisco. The wedge for free readers: embedded execution within the application process with S3-compatible storage backend ships under Apache 2.0, the only catalog free pick where storage decouples from compute on a columnar Lance format optimized for analytics workloads.

OSS embedded ships free under Apache 2.0 with embedded serverless execution and Arrow integration. Cloud free trial ships hosted managed LanceDB with S3-compatible storage backend and pay-per-query pricing. Enterprise ships custom pricing with BYOC option. Most analytics teams stay on OSS embedded indefinitely; Cloud trial is the path when managed S3 backend becomes load-bearing.

The trade-off versus Chroma is integration model; LanceDB embeds with S3 backend where Chroma is single-node in-memory. The trade-off versus Qdrant is production posture; LanceDB targets analytics batch workloads where Qdrant targets continuous query workloads. For embedded analytics teams with offline batch processing, LanceDB OSS is the right call.

Pros

  • Embedded execution within application process under Apache 2.0
  • S3-compatible storage backend decouples storage from compute
  • Built on Apache Arrow for analytics integration with pandas and DuckDB
  • Python, JS, and Rust SDKs for cross-language analytics workflows
  • Founded 2022 with strong columnar-format positioning for offline batch RAG

Cons

  • Limited multi-tenant support without built-in tenant scoping
  • Columnar batch architecture less optimal for continuous-query production RAG
OSS embedded freeCloud trial creditsPay-per-queryOSS Apache 2.0 free; Cloud trial available

Best for: Embedded analytics teams that want serverless vector storage with S3-compatible backend for offline batch RAG workflows.

OSS license & sovereignty
9
Query performance
9
Setup complexity
8
Value
9
Support
7
#2

Weaviate

5.3/10$240/yr more

Best free hybrid search OSS, vector plus keyword with GraphQL

Hybrid search OSS with GraphQL API and BSD-3 self-hosting plus managed cloud.

PlanMonthlyAnnualWhat you get
OSS self-hostedFreeBSD-3 licensed self-hosting with multi-vector and hybrid search.
Sandbox (Cloud)FreeFree 14-day trial of single sandbox cluster with all features unlocked.
Standard (Cloud)$25.00/mo$300.00/yr$25 monthly entry with pay-per-dimension and built-in hybrid search.
EnterpriseCustomCustomCustom pricing with BYOC option, multi-region, and premium support.

Weaviate is the hybrid search free pick and the right call for teams that need vector plus keyword search in one query. Founded 2019 in Amsterdam. The wedge for free readers: hybrid (vector plus BM25 keyword) search ships under BSD-3 self-host with GraphQL API and REST, the only catalog free pick where hybrid retrieval combines in one query rather than requiring application-level merging of separate searches.

OSS self-hosted ships free under BSD-3 with multi-vector and hybrid search plus GraphQL and REST. Sandbox cloud ships free fourteen-day trial with all features unlocked. Standard cloud is the upgrade tier at twenty-five dollars monthly starting with pay-per-dimension and hybrid search built in. Enterprise ships custom pricing with BYOC option. Most teams self-host BSD-3 indefinitely; Standard cloud is the upgrade trigger when SRE capacity becomes a bottleneck.

The trade-off versus Qdrant is OSS license posture; Weaviate is BSD-3 where Qdrant is Apache 2.0. The trade-off versus pgvector is integration breadth; Weaviate is standalone with GraphQL where pgvector lives in Postgres. For teams that need hybrid vector and keyword search, Weaviate OSS is the right call.

Pros

  • Hybrid vector and BM25 keyword search built in for retrieval quality lift
  • GraphQL API plus REST for JavaScript-first teams and modern app integration
  • BSD-3 OSS self-host with no licensing cost for unlimited deployment
  • Sandbox cloud fourteen-day trial with all features unlocked for evaluation
  • Founded 2019 in Amsterdam with strong enterprise traction

Cons

  • BSD-3 license posture differs from Apache 2.0 picks for some procurement teams
  • Pay-per-dimension cloud pricing complex for budgeting at scale
OSS BSD-3 freeSandbox 14d trialStandard $25/mo upgradeOSS BSD-3 free; 14-day cloud sandbox

Best for: Teams that need hybrid vector and keyword search in one query with GraphQL API and BSD-3 self-host.

OSS license & sovereignty
9
Query performance
8
Setup complexity
8
Value
9
Support
8
#3

Qdrant

4.8/10$240/yr more

Best free Rust-based production, filter-aware HNSW indexing

Rust-based vector DB with fastest filter-aware HNSW indexing under Apache 2.0.

PlanMonthlyAnnualWhat you get
OSS self-hostedFreeApache 2.0 Rust-based vector DB with filter-aware HNSW indexing.
Free Cluster (Cloud)Free1GB always-free production-ready cluster with low replication.
Standard (Cloud)$25.00/mo$300.00/yr$25 monthly for 4GB at $0.108 per GB-hour with higher availability.
Enterprise / HybridCustomCustomBYOC option with multi-region, SSO, and premium support.

Qdrant is the production-OSS free pick and the right call for production teams that want Rust-based performance without paying for managed cloud. Founded 2021 in Berlin. The wedge for free readers: Rust-based engine with filter-aware HNSW indexing maintains sub-millisecond latency under filtered queries, the only catalog free pick that delivers production-grade performance on customer infrastructure with no licensing cost.

OSS self-hosted ships free under Apache 2.0 with Rust-based engine and filter-aware HNSW. Free Cluster cloud ships always-free production-ready with one gigabyte storage. Standard cloud upgrade ships at twenty-five dollars monthly for four gigabytes with higher availability and replication. Most production teams land on Free Cluster for evaluation then pick OSS self-host or Standard cloud based on SRE capacity.

The trade-off versus Chroma is integration friction; Qdrant requires a long-running database server where Chroma embeds in the application. The trade-off versus pgvector is integration with existing data; Qdrant is standalone where pgvector lives in Postgres. For production teams that want Rust performance under Apache 2.0, Qdrant OSS or Free Cluster is the right call.

Pros

  • Rust-based engine for production-grade performance on customer infrastructure
  • Filter-aware HNSW maintains sub-millisecond latency under metadata-filtered queries
  • Apache 2.0 OSS self-host with no licensing cost for unlimited scale
  • Free Cluster cloud always-free production-ready with one gigabyte storage
  • Founded 2021 in Berlin with strong production-ready positioning

Cons

  • Smaller ecosystem than Pinecone or Weaviate for tooling integrations
  • No GraphQL API; REST and gRPC only for application integration
OSS Apache 2.0 freeFree Cluster 1GBStandard $25/mo upgradeOSS Apache 2.0 free; Free Cluster always-free

Best for: Production teams that want Rust-based performance under Apache 2.0 with filter-aware indexing for metadata-heavy RAG.

OSS license & sovereignty
9
Query performance
10
Setup complexity
8
Value
9
Support
7
#4

pgvector (Postgres)

4.6/10$168/yr more

Best free Postgres extension, vector search inside existing Postgres

Postgres extension adding vector search to Postgres on Supabase, Neon, RDS.

PlanMonthlyAnnualWhat you get
OSS extensionFreePostgres extension with HNSW and IVFFlat indexes for hybrid SQL plus vector queries.
Supabase FreeFree500MB Postgres with pgvector for up to ~5M small vectors free permanently.
Neon Launch$19.00/mo$228.00/yr$19 monthly with 10GB autoscaling Postgres and pgvector built in.
AWS RDS / AuroraFree$0.00/yrStandard RDS pricing with pgvector preinstalled for combined relational data.

pgvector is the Postgres-bundled free pick and the right call for teams already running Postgres. Open-source under Apache 2.0 since 2021, contributed by Andrew Kane. The wedge for free readers: vector search adds to the Postgres database your team already runs, the only catalog free pick where vector search lives alongside relational data with no new database, no new query language, and no new operational overhead.

OSS extension ships free as a Postgres extension on any deployment. Supabase Free covers five hundred megabyte Postgres with pgvector for up to around five million small vectors. Neon Launch is the upgrade tier at nineteen dollars monthly with ten gigabyte autoscaling Postgres and pgvector built in. AWS RDS and Aurora ship pgvector preinstalled at standard RDS pricing. Most Postgres teams stay on Supabase Free or self-hosted Postgres until vector volume crosses around fifty million vectors.

The trade-off versus Chroma is integration model; pgvector lives in Postgres where Chroma is embedded in application. The trade-off versus Qdrant is performance ceiling; pgvector handles up to around fifty million vectors comfortably while Qdrant scales further. For teams already on Postgres, pgvector is the right call.

Pros

  • Adds vector search to existing Postgres deployment with zero new infrastructure
  • Hybrid SQL plus vector queries combine in one database for relational-RAG patterns
  • Free on Supabase 500MB tier with around five million small vectors
  • Neon Launch upgrade at nineteen dollars monthly is cheapest paid pgvector path
  • Apache 2.0 OSS with broad ecosystem and mature production deployments

Cons

  • Scale ceiling around fifty million vectors; dedicated DBs win at higher scale
  • No native managed-vector-DB features like serverless or multi-region replication
OSS extension freeSupabase free 500MBNeon $19/mo upgradeOSS Apache 2.0 free; Supabase free permanent

Best for: Teams already running Postgres who want vector search added to the same database without operational overhead of a second DB.

OSS license & sovereignty
9
Query performance
8
Setup complexity
9
Value
10
Support
8
#5

Chroma

4.2/10$240/yr more

Best free for prototyping, Python-first single-node embedded

Lightweight Python-first embedded vector DB with single-node OSS plus managed cloud.

PlanMonthlyAnnualWhat you get
OSS self-hostedFreeApache 2.0 single-node embedded vector DB with Python and JS SDKs.
CloudFreeFree trial credits with pay-as-you-go pricing on hosted Chroma.
Cloud Standard$25.00/mo$300.00/yr$25 monthly starter with persistent collections and tenant scoping.
EnterpriseCustomCustomCustom pricing with on-prem deployment and dedicated support.

Chroma is the prototyping free pick and the right call for Python-first developers iterating on RAG pipelines in Jupyter notebooks. Founded 2022 in San Francisco and backed by a16z. The wedge for free readers: single-node embedded execution under Apache 2.0 ships with Python and JS SDKs and integrates natively with LangChain, LlamaIndex, and CrewAI, the lowest-friction free option for getting started with vector search.

OSS self-hosted ships free under Apache 2.0 with single-node embedded execution. Cloud free trial covers initial credits with pay-as-you-go pricing on hosted Chroma. Cloud Standard is the upgrade tier at twenty-five dollars monthly with persistent collections and multi-tenant scoping. Most prototyping teams stay on OSS embedded indefinitely; Cloud Standard is the upgrade trigger when production persistence becomes load-bearing.

The trade-off versus Qdrant is production scale; Chroma's single-node architecture limits scaling beyond around ten million vectors. The trade-off versus pgvector is integration depth; Chroma is standalone where pgvector lives in the Postgres database. For Python-first prototyping teams, Chroma OSS embedded is the right call.

Pros

  • Lowest-friction Python-first developer experience for prototyping
  • Single-node embedded execution under Apache 2.0
  • LangChain, LlamaIndex, CrewAI integrations work natively
  • Cloud Standard upgrade at twenty-five dollars monthly when production persistence matters
  • Founded 2022 with strong a16z backing and rapid iteration

Cons

  • Single-node architecture limits scale beyond around ten million vectors
  • No GPU acceleration; not the right fit for high-throughput production
OSS embedded freeCloud Standard $25/moPython-first SDKOSS Apache 2.0 free; Cloud trial credits

Best for: Python-first developers prototyping RAG pipelines in Jupyter notebooks who need the lowest-friction free vector DB.

OSS license & sovereignty
8
Query performance
7
Setup complexity
10
Value
9
Support
7

How we picked

Each pick gets a transparent composite score from price, features, free-tier availability, and editor fit. Pricing flows from our live database, so when a vendor changes prices the score updates here too.

We weight price at 40 percent, features at 30, free tier at 15, fit at 15. Chroma leads because the single-node embedded architecture matches prototyping launch shape and Python-first developer experience removes integration friction. See the parent /best/vector-databases guide for managed-cloud-only picks excluded from this lens.

We don't claim "30,000 hours of testing." Our methodology is the formula above plus the editor's published verdict for each pick. Verifiable, auditable, and updated when the underlying data changes.

Why trust Subrupt

We're a subscription tracker first, a buying guide second. Every claim on this page is something you can check.

By use case

Best free for prototyping and local dev

LanceDB

Read the full review →

Best free Rust-based production

Qdrant

Read the full review →

Best free Postgres extension

pgvector (Postgres)

Read the full review →

Best free hybrid search OSS

Weaviate

Read the full review →

How to choose your Free Vector Database

OSS license posture varies across the free vector DB lineup

The five catalog free picks ship under different OSS licenses. Chroma, Qdrant, pgvector, and LanceDB ship under Apache 2.0, the most permissive license preferred by enterprise procurement teams. Weaviate ships under BSD-3, similarly permissive but with subtle license-header requirements. Pinecone is closed-source SaaS and is excluded from this lens. Milvus is Apache 2.0 but operational complexity makes it less suitable for free-tier evaluation; teams reach Milvus when distributed multi-shard scale becomes load-bearing. The decision pivots on procurement license preference plus workload shape.

When does pgvector beat dedicated vector DBs?

pgvector is the right call when the team already runs Postgres and the relational data model carries metadata that RAG workloads need to filter on. Most production RAG workloads under ten million vectors run fine on pgvector with HNSW or IVFFlat indexes; vector-DB specialization pays off only at higher scale. The break-even depends on team operational capacity. Teams with mature Postgres expertise and existing PgBouncer plus replication infrastructure benefit from pgvector indefinitely. Teams with high-cardinality metadata filters or extreme query throughput hit Qdrant filter-aware HNSW or Weaviate hybrid search at lower scale than vanilla pgvector handles.

Chroma vs LanceDB for embedded use cases

Chroma and LanceDB both ship embedded vector storage but optimize for different workloads. Chroma is single-node Python-first with in-memory or local file persistence, optimal for prototyping in Jupyter notebooks and small production workloads where the database is queried continuously. LanceDB ships columnar Lance format with S3-compatible storage backend optimal for analytics batch workloads where the database is queried periodically rather than continuously. The decision pivots on whether the team needs continuous-query RAG (Chroma) or periodic-batch analytics on embeddings (LanceDB).

Hybrid search: Weaviate plus BM25 vs Qdrant filter-aware HNSW

Hybrid retrieval matters for RAG quality on mixed text-and-metadata data. Weaviate ships hybrid (vector plus BM25 keyword) search natively in one query, combining semantic similarity with keyword matching for retrieval quality lift. Qdrant ships filter-aware HNSW indexing that maintains sub-millisecond latency when queries combine vector similarity with metadata filters. The two pick different sides of the hybrid coin. Weaviate fits teams where keyword precision matters (legal docs, technical documentation, structured chunks). Qdrant fits teams where metadata filtering matters (multi-tenant RAG, time-bound retrieval, source-filtered queries). pgvector handles both at smaller scale via SQL plus vector hybrid queries.

When to upgrade past free to managed cloud (cross-link to parent)

Free OSS paths cover most evaluation and small-team production workloads but each pick has clear upgrade triggers. Chroma OSS outgrows past around ten million vectors; Cloud Standard at twenty-five dollars monthly is the upgrade. Qdrant OSS outgrows when SRE capacity becomes a bottleneck; Free Cluster covers one gigabyte free or Standard at twenty-five dollars monthly is the upgrade. pgvector outgrows past around fifty million vectors; dedicated vector DBs win at higher scale. Weaviate OSS outgrows when SRE capacity becomes a bottleneck; Standard at twenty-five dollars monthly is the upgrade. LanceDB OSS outgrows when managed S3 backend becomes load-bearing; Cloud trial is the path. At any of those triggers, see [our /best/vector-databases guide](/best/vector-databases) for the full lineup including Pinecone serverless and Milvus distributed.

Frequently asked questions

Which free vector database is the most generous?

It depends on workload shape. Chroma OSS is most generous on prototyping ergonomics. Qdrant OSS is most generous on production performance with Rust-based filter-aware HNSW. pgvector is most generous on integration cost for teams on Postgres. Weaviate OSS is most generous on retrieval features with hybrid search. LanceDB OSS is most generous on analytics workflows with S3 backend.

Can I run a production RAG workload on a free OSS vector database?

Yes for most workloads. Most production RAG under 10M vectors runs fine on pgvector or Chroma. Production RAG up to 100M vectors with metadata filtering runs fine on self-hosted Qdrant. Hybrid vector and keyword search runs on self-hosted Weaviate. Trade-off is operational responsibility; OSS self-host requires SRE capacity for backups and monitoring.

Should I use Pinecone or self-host Qdrant for a free vector DB?

Pinecone Starter is closed-source SaaS with 5 free indexes plus monthly read and write limits. Qdrant OSS is Apache 2.0 self-host with unlimited scale on customer infrastructure. Teams without SRE capacity pick Pinecone Starter for zero-operations. Teams with mature Kubernetes infrastructure pick Qdrant OSS for unlimited scale. Most production teams that scale past Pinecone Starter pricing migrate to Qdrant or Weaviate self-host within the first year.

Does Subrupt earn a commission from these free picks?

On most paid upgrades. We disclose this on every /best page. OSS self-hosted free tiers themselves have no transaction. Paid managed-cloud upgrades on Chroma, Qdrant, Weaviate, and LanceDB have plans where we earn commission only on conversion. The composite ranking weights price at 40 percent, features at 30, free tier at 15, fit at 15; none tuned by affiliate rate.

Why is Chroma ranked first over the more performant Qdrant?

Chroma wins on prototyping launch shape because Python-first single-node embedded matches the Jupyter-notebook RAG iteration pattern better than production-first Qdrant. Qdrant is genuinely more performant for production workloads but the workload shape Qdrant serves is continuous-query production RAG while Chroma serves prototype iteration. Prototyping teams lean Chroma; production teams lean Qdrant.

Can I migrate vector data between OSS vector DBs?

Yes within a few hours of code. Vector data plus metadata exports and imports between any two vector DBs given matching dimension counts. Chroma to Qdrant migration is a Python script reading collection items and inserting into Qdrant collections. pgvector to Qdrant is a SQL export plus Python import. Most production RAG teams do not migrate often because the cost is low; the lock-in concern that some incumbent guides emphasize is overstated.

How does pgvector handle large-scale vectors?

pgvector with HNSW indexing handles up to around 50M vectors comfortably on a single Postgres instance. Beyond that scale, query latency degrades and dedicated vector DBs deliver better performance. Partitioning plus read replicas extend the scale ceiling but require more operational overhead. For teams running multi-terabyte Postgres with mature SRE expertise, 100M vectors is achievable but not the typical recommended path.

What about Milvus on the free tier?

Milvus OSS ships under Apache 2.0 with distributed architecture and GPU acceleration but operational complexity makes it less suitable for free-tier evaluation. Self-hosting at production scale requires SRE expertise plus Kubernetes. Zilliz Cloud Free at 5GB covers small evaluation; Standard starts at $65/mo. Milvus reaches its strength at 1B vectors plus. For typical free-tier RAG evaluation, Chroma or Qdrant fit better.

EU data residency: which free OSS picks fit GDPR-strict deployments?

All five OSS picks ship full data residency control through self-hosting. Chroma, Qdrant, pgvector, Weaviate, and LanceDB all run on customer infrastructure in any jurisdiction. Managed cloud paths for Qdrant, Weaviate, and Chroma ship EU regions on paid tiers. For default EU residency without contract negotiation, OSS self-host on EU infrastructure or Weaviate Cloud EU region are the cleanest catalog fits.

How often is this guide updated?

We re-review pricing and features annually at minimum, with mid-year refreshes when major vendor announcements happen. Chroma launched the Cloud product 2024. Qdrant Free Cluster launched 2023 and remains always-free. pgvector has shipped consistent Apache 2.0 since 2021. Weaviate Cloud Standard at twenty-five dollars monthly since 2023. LanceDB launched Cloud product 2024. The lastReviewed date reflects the most recent editorial pass.

Subrupt Editorial

The team behind subrupt.com. We track subscriptions, surface cheaper alternatives, and publish buying guides where the score formula is on the page so you can recompute it yourself. We do not claim 30,000 hours of testing. What we claim is live pricing from our database, a transparent composite score, and honest savings math against a category baseline.

Last reviewed

Citations

Affiliate disclosure: Subrupt earns a commission when you switch to a service through our recommendation links. This never changes the price you pay. We only recommend services where there's a real cost or feature advantage for you, and our picks are based on the data on this page, not on which programs pay the most.

Related buying guides

Track your subscriptions on Subrupt

Add the Free Vector Database you pay for and see how much you'd save by switching.

Open dashboard

More buying guides

Independent rankings for the subscriptions worth paying for.

See all guides