Cheaper than the search and vector platforms we compared
Price and performance against the search and vector platforms people ask us about. Every number links to its source.
Monthly cost
List prices in US East for two workloads. Open a platform for the line items: storage, writes, search, data returned and minimums.
| Small RAG | Catalog | |
|---|---|---|
| WarpSearch | $10 | $139 |
| Aiven for OpenSearch | $365 to $730 | $730 to $1,455 |
| Pinecone | $118 to $137 | $8,100 to $9,200 |
| turbopuffer | $16 | $536 to $556 |
| Algolia (Grow) | About $27,000 (keyword only) |
| 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. |
Blank means the platform does not fit the workload. Ranges cover regions, plan choices, or how the vendor sizes compute.
Where each one wins
| Platform | Bottom line |
|---|---|
| Aiven | If your traffic is not steady around the clock, WarpSearch costs a fraction of a fixed cluster. Aiven makes sense when you need full OpenSearch control and steady load. |
| Pinecone | For RAG and product search, WarpSearch is 10x to 60x cheaper than Pinecone on demand and adds full keyword search and facets. Pinecone is the proven choice at billion-vector scale. |
| turbopuffer | Same architecture, different meter. WarpSearch charges per request; turbopuffer charges per GB searched. The more you query a large index, the more WarpSearch saves. For a small, rarely queried index, the two cost about the same. |
What we measured
| Test | Result |
|---|---|
| Keyword search, 2.8M docs (LoTTE) | 2.1 ms median, 3,700 queries/s |
| Keyword search, 10M passages (MS MARCO v2) | 2.4 ms median (5.4 ms with an exact total hit count) |
| Keyword search, small BEIR sets | 1 to 6 ms median |
| Vector search, 10M x 128-dim (BigANN SIFT) | 0.964 recall@10 at 6.2 ms median, 162 queries/s |
| Vector search, 1M x 1536-dim (arXiv) | 0.961 recall@10 at 15.7 ms median |
| Filtered vector search, 10M (YFCC) | 0.959 recall@10 at 10.4 ms median |
| Log filters, 22M events (http_logs) | 1 to 2 ms median |
| Log aggregations, 22M events | 190 to 430 ms median (sort, daily histogram, avg by status) |
| Ingest, short text | 63,000 to 97,000 docs/s |
| Ingest, logs (247M events) | 76,000 to 81,000 events/s |
| Ingest, 768 to 1536-dim vectors | 2,200 to 4,000 vectors/s |
| Time to searchable | Seconds, or immediate with commit=true |
| Cold start | Not measured yet. First query on a 2.3 GB index: 17 to 19 ms vs 2.7 ms warm (page cache only) |
Measured October 2026 through our Elasticsearch-compatible API, warm cache, client in the same region, single runs. Every test ran on deliberately minimal hardware, 2 to 4 CPUs (a 4 CPU container or an AWS i8g.large with 2 vCPUs), so these are the least favorable results we can produce. More hardware only improves them. Numbers exclude internet round trips. We did not run other platforms ourselves. Each platform page shows what that vendor publishes, with links.
Limitations of each platform
What each one cannot do, or does only in part, taken from the vendor's own pricing and documentation.
| Platform | Limitations |
|---|---|
| Algolia (Grow) | Grow is keyword search only. Semantic search (NeuralSearch) is on the Elevate plan, which needs an annual contract and custom pricing, so this is the keyword-only price. Priced at $0.50 per 1,000 searches and $0.40 per 1,000 records above the free quota. |
| Aiven | Fixed monthly cluster plans, no scale to zero: you pay for every hour whether you query or not, and you size nodes, disk and shards yourself. |
| Pinecone | Vector database with no aggregations or facets. Hybrid search needs sparse vectors (scores you normalize yourself) or fusion on your side. Query cost grows with namespace size. |
| turbopuffer | Own API, not Elasticsearch compatible. Aggregations are capped at 10k groups, writes allow at most one commit per second per namespace, and cold queries are slow (874 ms median on 10M vectors). Query cost grows with namespace size. It does support BM25 plus vector search with server-side RRF. |
Prices checked October 20, 2026. List prices in US dollars, before discounts. See WarpSearch pricing. Found something out of date? Tell us.