Use cases / Vector database alternative
A vector database that also does keyword search.
Most vector databases make you bolt on a search engine for exact matches, or settle for vectors only. WarpSearch stores vectors with your text and metadata in one index, filters during the search, and fuses keyword and vector results in one query.
- 6.2 msmedian at 0.964 recall@10 on 10M vectors, on 2 vCPUs
- 10.4 msmedian for filtered vector search at 0.959 recall@10 on 10M vectors
- $10a month for 1M chunks and 1M queries
Everything around the vectors
turbopuffer-compatible
Writes, queries and namespaces through the turbopuffer API, so existing SDK code points at a new URL. See turbopuffer compatibility.
Filtered search that keeps recall
Filters apply inside the approximate search, so a narrow filter still returns k good results.
Smaller, cheaper vectors
float16, bfloat16, int8 and bit encodings, plus server-side Matryoshka truncation. See Dimensions and storage.
Several vectors per document
Store chunk or image vectors on one document and score it by its best match. See Multi-vector fields.
Hybrid when you need it
Add BM25 to the same query and fuse the lists with RRF. No second system.
A namespace per tenant
Idle namespaces cost only their storage, so one per user or customer stays cheap.
Vector database pricing compared
Monthly cost at list prices in US East. Open a platform for the line items and sources.
| Platform | Small RAG | Catalog |
|---|---|---|
| WarpSearch | $10 | $139 |
| Pinecone | $118 to $13712 to 14x cheaper | $8,100 to $9,20058 to 66x cheaper |
| turbopuffer | $161.6x cheaper | $536 to $5563.9 to 4x cheaper |
| S3 Vectors | Vector search only, no hybrid | About $9616.9x cheaper |
| OpenSearch Serverless | $357 to $53036 to 53x cheaper | $550 to $1,0804 to 7.8x cheaper |
| Elastic Serverless | About $100 to $21010 to 21x cheaper | $210+1.5x cheaper or more |
| Small RAG | 1M text chunks with 768-dim embeddings, 7 GB of JSON. 1M queries a month spread through the day. 1 GB written a month. |
|---|---|
| Catalog | 5M products with 384-dim embeddings, 10 GB of JSON. 50M queries a month (about 19 a second on average). 20 GB written a month. |
List prices in US dollars, checked October 20, 2026. Ranges cover regions, plan choices, or how the vendor sizes compute. All comparisons and sources.
Good to know
Approximate search uses IVF with PQ quantization, not HNSW. You compute embeddings yourself, and sparse vectors are in development.
Read next
Free up to 1 GB, no card. Pay per GB and per request after that.