- Tech Lead · AWS OpenSearch
infino authors --list
02 Vinay Kakade
- Head of Eng · AWS OpenSearch
03 Asif Makhani
- GM, AWS OpenSearch
- Chief Architect · AWS OpenSearch
infino cost
Estimate the cost of your workload on Parquet.
storage $0.30 per GB-month
$3 write tokens $0.15 per 1k WT
i
Write tokens measure ingest work. Batched writes assumed. methodology →
$2 read tokens $0.05 per 1k RT
i
Read tokens measure query work. Pinned tables assumed at higher volumes. methodology →
$70 returned bytes $0.05 per GB returned
i
Query result payload sent back to the client. methodology →
$1 infino $75/month account minimum
Elasticsearch $1,314
Elastic Cloud Hosted · Platinum
data nodes2 nodes · 2 zones × 8 GB · $0.09/GB-hr$1,051
Kibana2 × 2 GB$263
master quorumdata nodes + included 1 GB tiebreaker$0
total$1,314
methodology → 17.4× more OpenSearch $1,751
OpenSearch Service
data nodes3 × r6g.2xlarge.search · $0.669/hr$1,465
masters3 × m6g.large.search$280
EBS gp344 GB$5
total$1,751
methodology → 23.2× more Infino Cloud Pricing. methodology →
infino bench --latency
See the latest benchmarks.
Workload Cohere, 768 dimensions, 10M docs, top-10
p50
5ms 314ms
p99
12ms 850ms
- Warm cache
- Cold cache
Workload Zipfian corpus, BM25, 10M docs, median query shape
p50
2ms 275ms
p99
7ms 720ms
- Warm cache
- Cold cache
Workload 10M rows · warm p50 · bounded results
metadata aggregate 0.3ms
lookup aggregate 3ms
scan aggregate 41ms
crosstab aggregate 75ms
infino compatibility
See how it fits into your existing stack.
query DSL
elasticsearch opensearch
parquet
snowflake databricks
catalog integration
apache iceberg apache hudi
cat query.sql
Accurate, token-saving, & expressive querying.
Table-valued functions let search results compose as ordinary SQL tables, so one query replaces several. Agents make fewer LLM round trips; humans find what they need faster.
-- one query: hybrid search + SQL filters SELECT _id, title, score FROM hybrid_search( -- FTS + vector, fused by RRF 'logs', 'body', 'disk full', -- the text side 'embedding', :q, 50 -- the vector side, top 50 ) WHERE level = 'error' -- pushed-down filter AND ts > now() - interval '24 hours' -- on the same pass ORDER BY score DESC -- one fused ranking LIMIT 10;
cat analytics.sql
Analytics & Search Combined.
Search results are a relation, so follow-up questions stay in SQL, inline in the same query. Getting from “find disk-full errors” to “which team had them” takes one query, for a person or an agent.
-- a ranked search is a relation: join it, group it, aggregate it SELECT s.team, count(*) AS hits, -- rows per team avg(h.score) AS relevance -- mean fused score FROM hybrid_search( -- the same hybrid search 'logs', 'body', 'disk full', 'embedding', :q, 1000 ) AS h JOIN services s ON s.id = h.service_id -- a plain table WHERE h.ts > now() - interval '7 days' GROUP BY s.team ORDER BY hits DESC;
infino spec
- search
- full-text · vector · hybrid
- index
- BM25 (PFOR-delta, FST) · HNSW · OPANN + Sq16
- engine
- Rust
- language
- SQL (Apache DataFusion) · REST · Query DSL
- storage
- object storage, S3 · GCS · Azure Blob · on-prem
- format
- Apache Parquet
- deploy
- on-prem · self-hosted cloud · hosted cloud
- core license
- Apache-2.0
infino init