Serverless Elasticsearch
Serverless Elasticsearch. No VCUs.
WarpSearch serves the Elasticsearch API as a serverless service. Point your Elasticsearch client, Logstash or Beats at api.warpsearch.io and index. No VCUs, nodes or shards to size, and no search baseline running while nobody searches. You pay per GB and per request, from $10 a month, with a 99.9% uptime SLA on every paid account.
- 0 VCUsNo baseline and no VCUs per query to estimate. Billing is per GB stored, per GB written and per request.
- $10a month for a small app. Elastic's own examples put a small Elastic Cloud Serverless search project at about $100 to $210 a month.
- 2 to 6 msmedian keyword search on 2.8M to 10M documents, measured on 2 to 4 CPUs. Elastic publishes 22 to 67 ms p99 for Elastic Cloud Serverless.
Your Elasticsearch client, a new host
# pip install elasticsearch from elasticsearch import Elasticsearch es = Elasticsearch("https://api.warpsearch.io/compat/es", api_key="ws_...") es.indices.create(index="products", mappings=mapping) es.bulk(operations=actions, refresh=True) es.search(index="products", query={"match": {"title": "wireless headphones"}})
Up and running in three steps
Create an API key
Sign up and create a key in the console. Keys have roles: viewer searches, writer also indexes, editor also creates indices, admin also deletes them. Pass it as
api_keyin the client, or asAuthorization: Bearer,ApiKeyorX-API-Key.Create your index with a mapping
PUT /productswith your mappings and analysis settings. Multi-fields liketitle.keyword,copy_toand language analyzers are supported. There is no dynamic mapping, so send the mapping first. See Schemas and fields.Point your client or shipper at the endpoint
Base URL
https://api.warpsearch.io/compat/es, port 443. Logstash uses the elasticsearch output with the key incustom_headers; Beats setpath: /compat/esand a header. Bulk load, call_refreshonce, search. Free up to 1 GB with no card.
Works with the Elasticsearch tools you run
Every official client takes a base path and an API key. Settings for each are in the connection table.
- elasticsearchPython
- @elastic/elasticsearchJavaScript
- Java API clientRestClient or Rest5Client
- go-elasticsearchGo
- Elastic.Clients.Elasticsearch.NET
- elasticsearchRuby, PHP, Rust
- Logstashelasticsearch output
- Filebeat and Beatstested with Filebeat 8.11
- Fluent Bites output
- Vectorelasticsearch sink
- GrafanaElasticsearch data source
- curlREST
Hosted Elasticsearch with nothing to run
A request is a request
$2.00 per million requests, $0.08 per GB stored a month and $1.00 per GB written. You can work out the bill before you get it. Elastic Cloud Serverless bills VCU-hours and does not publish how many VCUs a query uses.
No nodes, shards or VCU baseline
Nothing to size and no search tier billed while nobody searches. A collection grows with your data and your traffic, and an idle one costs only its storage.
Keyword, vector and hybrid in one index
dense_vector fields up to 8192 dimensions, top-level knn with filters applied inside the search, and rrf and linear retrievers for hybrid ranking. See Hybrid search.
Logs from Beats and Logstash
Ship with Filebeat, Logstash, Fluent Bit or Vector, keep @timestamp and event.created names in queries, and read it all in Grafana. Storage from $0.08 down to $0.035 per GB a month at volume. See Log analytics.
99.9% uptime SLA from $10
Every paid account has it, with no second zone or higher tier to buy.
Any cloud, or your own
AWS, Google Cloud or Azure regions close to your app, or WarpSearch inside your own account with the same APIs. Ask about BYOC.
How fast is it next to Elastic Cloud Serverless?
Elastic publishes p99 latency from its own Rally runs on Serverless, without stating the hardware. Our numbers are medians on 2 to 4 CPUs over different datasets, so treat the comparison as indicative, not like for like.
| WarpSearch | Elastic Serverless | |
|---|---|---|
| Keyword search | 1 to 6 ms median (2.1 ms on 2.8M docs, 2.4 ms on 10M passages) | 35 ms p99 on Rally wikipedia |
| Vector search | 6.2 ms median at 0.964 recall@10 on 10M vectors, 15.7 ms at 1536 dims | 22 ms p99 on BEIR NQ, 1536 dims |
| Ingest | 63,000 to 97,000 short documents a second | Not published for Serverless |
| Time to searchable | Immediate with refresh=true or _refresh. Otherwise within a few minutes | 5 seconds by default |
| Idle | Storage only. A collection idle for 15 minutes reloads on its next request; optional keep-warm keeps it ready. We have not published cold-start timings yet | Search keeps a baseline, billed at a reduced idle rate |
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.
What serverless Elasticsearch costs for two workloads
Monthly cost at list prices in US East for two common search workloads. Open a platform for the line items and sources.
| Platform | Small RAG | Catalog |
|---|---|---|
| WarpSearch | $10 | $149 |
| Elastic Serverless | About $100 to $21010 to 21x cheaper | $210+1.4x cheaper or more |
| Elastic Cloud | $93 to $2859.3 to 29x cheaper | $516 to $1,5003.5 to 10x cheaper |
| OpenSearch Serverless | $357 to $53036 to 53x cheaper | $550 to $1,0803.7 to 7.2x cheaper |
| OpenSearch Service | $132 to $39713 to 40x cheaper | $400 to $1,5752.7 to 11x cheaper |
| Aiven | $365 to $73037 to 73x cheaper | $730 to $1,4554.9 to 9.8x cheaper |
| Pinecone | $118 to $13712 to 14x cheaper | $8,100 to $9,20054 to 62x cheaper |
| turbopuffer | $161.6x cheaper | $536 to $5563.6 to 3.7x cheaper |
| 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.
What serverless Elasticsearch includes
This is what the service includes today. A request for anything else answers 404 or 400 unsupported_feature.
| Area | Status | Notes |
|---|---|---|
Search: _search, _msearch, _count, Query DSL | Available | match, multi_match, combined_fields, bool, term, range, prefix, wildcard, regexp, fuzzy, function_score, query_string, geo queries, sort, collapse, post_filter. Scores differ from Elasticsearch |
| Aggregations | Most | terms, multi_terms, histograms, auto_date_histogram, ranges, filters, metrics, cardinality, percentiles, top_hits, geo grids. No pipeline aggregations, composite or scripts. Results do not parse through the Java API client; use the low-level client or raw JSON for aggregations |
| kNN, hybrid and retrievers | Available | Top-level knn, knn beside a query, rrf and linear retrievers. No query_vector_builder, semantic or sparse_vector |
| Highlighting | Available | Per-hit snippets, also in _msearch, top_hits and collapsed inner hits |
| Index and mapping APIs | Available | Create with mappings and analysis settings, get, delete, add fields, multi-fields and copy_to. No dynamic mapping, templates, aliases or data streams |
Documents and _bulk | Most | index, create, delete, doc_as_upsert, get, mget, refresh, forcemerge. No partial _update, update_by_query, delete_by_query, reindex or ingest pipelines |
| Clients and shippers | Available | All official clients, Logstash, Beats, Fluent Bit, Vector, Grafana. Shippers need template and ILM setup turned off |
| Authentication | API keys | api_key or a Bearer, ApiKey or X-API-Key header, with per-key roles. No HTTP Basic, no Elasticsearch id:api_key pairs |
Scroll, search_after, point in time | Not supported | Page with from and size, or a range filter on the last hit's sort values |
| Nested and join fields, scripts, runtime fields | Not supported | |
Kibana, ILM, Watcher, snapshots, _cat, _sql, ES|QL | Not supported | Grafana works for dashboards. SQL is on the WarpSearch API |
Full tables: Elasticsearch reference.
Questions
Is there a serverless Elasticsearch?
Yes. WarpSearch serves the Elasticsearch API with no nodes, shards or VCUs, billed per GB stored, per GB written and per request from $10 a month. Elastic Cloud Serverless is Elastic's own serverless offering, billed in VCU-hours for search projects.
Is WarpSearch Elasticsearch as a service?
It is a hosted, serverless service that serves the Elasticsearch API. You get an endpoint and an API key, not a deployment to size. Official clients, Logstash and Beats connect as they would to a cluster. Kibana, scroll and nested documents are not part of it; the table above lists the scope.
How is this different from Elastic Cloud Serverless?
Elastic Cloud Serverless bills $0.14 per ingest VCU-hour and $0.09 per search VCU-hour plus retention, keeps a search baseline running at an idle rate, and does not publish how many VCUs a query uses. WarpSearch bills each request at a published rate and an idle collection costs only its storage. See the line-by-line cost comparison.
Do Logstash and Beats work?
Yes. Logstash uses the elasticsearch output with hosts => ["https://api.warpsearch.io:443/compat/es"] and the key in custom_headers. Beats set path: /compat/es and a header, with template and ILM setup turned off. We tested Filebeat 8.11. The Elasticsearch reference has the settings for each shipper.
Does Kibana work?
No. WarpSearch serves the Elasticsearch REST API, not Kibana. Grafana's Elasticsearch data source works for logs and metrics panels. ILM, Watcher, ingest pipelines and ES|QL are also not supported; the table above has the full scope.
Does it support vector and hybrid search?
Yes. Map a dense_vector field, send a top-level knn with a filter, or combine keyword and vector results with rrf or linear retrievers. Vectors live in the same index as your text fields. See Hybrid search.
Elastic Cloud Hosted or serverless?
Elastic Cloud Hosted means sizing a cluster, from $99 a month, with Kibana on a node you pay for. Elastic Cloud Serverless removes the sizing but bills VCU-hours. WarpSearch has neither nodes nor VCUs, and starts at $10. See the Elastic Cloud Hosted comparison.
Is WarpSearch an alternative to Elasticsearch?
For search, logs and vector workloads that use the Elasticsearch API, yes, served serverless and billed per request. It does not replace Kibana, scroll exports or Painless scripts.
Read next
Free up to 1 GB, no card. Pay per GB and per request after that.