TurboPuffer alternative. Vector search on an open format you own.
Turbopuffer and LanceDB both run vector, full-text, and hybrid search at multi-billion scale. The difference is the format.
Turbopuffer data is reachable only through Turbopuffer's API, so anything beyond vectors is a second copy of data that lives elsewhere, kept in sync.
LanceDB stores everything in the open Lance format, in your own bucket. One copy, read by search, analytics, and any engine you already run.
.svg)
Why teams switch
One copy, no sync
turbopuffer's proprietary format means non-vector data is a duplicate of something that primarily lives elsewhere. Lance holds the source of truth, so search reads the same copy as everything else. No sync job, no drift.
Open format, queryable from anything
Lance is an open spec: file, table, and catalog. Query the same data from Spark, Trino, Ray, Python, and SQL. No data trapped behind one API.
Vector search at multi-billion scale
Vector, full-text, and SQL in one query. Character.AI moved full-text search off Elasticsearch onto LanceDB and cut P90 latency by over 90%.
One data store for everything
The same store also runs feature engineering, curation, training-data caching, and agentic search. Land on search; the open format means the rest is already in place when you want it.
Comparison
The Power of the Lance Format
Vector Search
- Fast scans and random access from the same table — no tradeoff
- Zero-copy access for high throughput without serialization overhead
Multi-Modal
- Raw data, embeddings, and metadata in one table — not pointers to blob storage
- No separate metadata store to keep in sync
Enterprise-Grade Requirements
Security
Granular RBAC, SSO integration, and VPC deployment options.
Governance
Data versioning and time-travel capabilities for auditability.
Support
Dedicated technical account management and guaranteed SLAs.
