Pricing for the AI-native database. Single-tenant compute. Predictable quotas. Opt-in capabilities.
Single-tenant compute, predictable per-call quotas, opt-in capabilities. Trial available on Whisper.
Four compute tiers.
Whisper
Prototype · hobby · side project
- + 1 vCPU · 2 GB RAM
- + ~50 req/s sustained
- + 500 Mbps network
- + 1 M DB calls / mo
- + 3 K natural-language Q / mo
- + 100 GB outbound / mo
One developer, a handful of rows, an agent that needs a memory. 1 M DB calls + 2.5 K natural-language queries a month.
Thunder
Startup · production · agent fleet
- + 4 vCPU · 16 GB RAM
- + ~2000 req/s sustained
- + 2.5 Gbps network
- + 10 M DB calls / mo
- + 25 K natural-language Q / mo
- + 500 GB outbound / mo
A working product and a room full of agents. 10 M DB calls + 25 K natural-language queries, milliseconds.
Storm
High-throughput · low-latency · SLA
- + 8 vCPU · 32 GB RAM
- + ~6000 req/s sustained
- + 10 Gbps network
- + 50 M DB calls / mo
- + 100 K natural-language Q / mo
- + 2 TB outbound / mo
Trading floors, sensor networks, call-center analytics. 50 M DB calls + 100 K natural-language queries, 99.95 % SLA.
Enterprise
Custom SLA · BYOK · dedicated support
- + 32 vCPU · 128 GB RAM
- + custom sustained
- + 25 Gbps network
- + negotiated DB-call quota
- + negotiated AI-query quota
- + negotiated outbound network
Regulated industries, multi-region deployments, custom retention. Negotiated quotas, customer-managed keys, named support engineer.
* storage billed separately (see below) · usage above the quota is metered at clear, listed rates
Add storage to the compute tier.
NVMe-backed EBS, AES-256 at rest, region-isolated. IOPS burst scales with the tier. Nightly snapshots are included in every storage tier; PITR is available on Storm compute.
* $0.30 / GB-month above 2 TB · contact sales for custom sizing
You only pay if you go past what's included.
Every tier ships with a generous monthly quota. Below the quota everything is included — no per-query fee, no token surcharge. Above the quota, three line items kick in at clear, listed rates: $0.02 per natural-language query, $0.0001 per DB call, and $0.12 per GB of outbound network. Each line is itemised on the invoice and can be capped from the console.
One English (or other-language) question sent to
/v1/ask.
Cached answers are free; only fresh compiles count.
Insert, get, query, watch — every direct DB-API hit. The quota covers a healthy amount of traffic; this rate kicks in only if your fleet outgrows the tier mid-month.
Bytes leaving the region. Cross-region replies count toward this; intra-AZ replies do not. Pass-through metered cleanly — never buried in the base price.
* regions outside the primary (currently Mumbai) carry a 15 % uplift on compute + storage
* you can cap monthly overage from the console — instance returns HTTP 402 once the cap is hit
Bolt on only what your workload uses.
The base tier covers the substrate every workload needs — CAS, SELECT, INSERT, DELETE, boolean full-text, sealed-segment PITR on Storm. Capabilities only some workloads need — JOINs, BM25, vectors, graph, transactions, sub-second PITR — are flat-priced opt-in add-ons. Toggle any of them from the console; the meter prorates to the day.
SQL Pro
JOINs · GROUP BY · OUTER · HAVING
Full analytics SQL on your tenant. GROUP BY with COUNT, SUM, AVG, MIN, MAX; INNER, LEFT, RIGHT, and FULL OUTER joins; HAVING filters; and chained joins across up to five tables. The base tier ships SELECT, INSERT, and DELETE — this turns the rest on.
good for Analytics dashboards, BI rollups, anything that needs a JOIN.
Add to planVector Search
HNSW · cosine · filtered topk · tunable speed/recall
HNSW vector index with metadata filters and tunable speed/recall. Default high_recall mode hits recall@10 = 0.96 at 100k with p99 109 ms; opt into fast mode for p99 37 ms (recall ≈ 0.69) when latency dominates. Cosine, dot, and L2 metrics; filtered topk via metadata equality; f32 SIMD distance kernels and a deserialised graph cache. Brute-force fallback for small indexes.
good for RAG pipelines, recommendation, embedding-heavy agent memory.
Add to planFull-Text Pro
BM25 · phrase · 18-language stemming
BM25 ranked search with phrase queries and Snowball stemming for 18 languages. Lucene-default scoring (k1=1.2, b=0.75), position-list intersection for phrase queries, and a UAX #29 Unicode tokenizer covering Latin, Cyrillic, CJK, Arabic, and Hindi. Optional analyzer pipeline with diacritics fold and stopwords.
good for Search bars, log search, multilingual content lookup.
Add to planGraph
BFS · weighted shortest path · Dijkstra
Forward and reverse traversal, BFS, and weighted shortest path with Dijkstra. One-hop neighbours in either direction, bounded BFS, path reachability, and caller-supplied weight functions. Reverse traversal works on self-relations via direction-tagged keys.
good for Social graphs, supply-chain reachability, dependency walks.
Add to planTransactions
multi-row snapshot isolation
Multi-row snapshot isolation with optimistic conflict detection. Atomic commits across rows, indexes, vectors, and full-text postings; begin, get, put, delete, commit, and abort semantics with deterministic conflict resolution under concurrent writers.
good for OLTP workloads that need to mutate multiple rows atomically.
Add to planSub-second PITR
per-record point-in-time recovery
Per-record point-in-time recovery. Restore to any timestamp with sub-second granularity via WAL frame v2 with embedded timestamp_micros and a continuous tail-shipper. Sealed-segment PITR remains included in the base tier on Storm.
good for Regulated workloads that need to rewind to a specific event.
Add to planMulti-Writer Cluster
active-active · cross-region
Cross-region active-active replication on Enterprise. Talk to sales for sizing — clusters are provisioned and tuned by the OriginChain team to match your latency and durability targets.
good for Multi-region tenants that need writes accepted in every region.
Contact sales
* add-ons attach to any compute tier · billed alongside compute + storage
* meter prorates to the day on enable and disable
The detailed comparison.
Rows marked Included ship in every tier. Rows naming an add-on (SQL Pro, Vector Search, Full-Text Pro, Graph, Transactions, Sub-second PITR, Multi-Writer Cluster) are opt-in capabilities that attach on top of any compute tier — see the add-ons section above for prices.
| Capability | Whisper $99/mo | Thunder $599/mo | Storm $1,499/mo | Enterprise from $4,999/mo |
|---|---|---|---|---|
| vCPUs | 1 | 4 | 8 | 32 |
| RAM | 2 GB | 16 GB | 32 GB | 128 GB |
| Network egress | 500 Mbps | 2.5 Gbps | 10 Gbps | 25 Gbps |
| Sustained req/s | ~50 | ~2 000 | ~6 000 | custom |
| Row capacity (soft) | 5 M | 250 M | 1 B | 10 B+ |
| DB calls / month | 1 M | 10 M | 50 M | negotiated |
| Natural-language Q / month | 2 500 | 25 000 | 100 000 | negotiated |
| Outbound network / month | 100 GB | 500 GB | 2 TB | negotiated |
| Concurrent watches | 20 | 1 000 | 5 000 | unlimited |
| SQL — SELECT / INSERT / DELETE | Included | Included | Included | Included |
| Single-row CAS | Included | Included | Included | Included |
| Boolean full-text search | Included | Included | Included | Included |
| SQL Pro (JOIN, GROUP BY, OUTER) | SQL Pro | SQL Pro | SQL Pro | SQL Pro |
| BM25 + phrase + stemming | Full-Text Pro | Full-Text Pro | Full-Text Pro | Full-Text Pro |
| Vector search (HNSW) | Vector Search | Vector Search | Vector Search | Vector Search |
| Graph traversal + Dijkstra | Graph | Graph | Graph | Graph |
| Multi-row transactions | Transactions | Transactions | Transactions | Transactions |
| Sealed-segment PITR | — | — | Included | Included |
| Sub-second PITR | — | Sub-second PITR | Sub-second PITR | Sub-second PITR |
| Multi-writer cluster | — | — | — | Multi-Writer Cluster |
| Backups | daily · 7d | 1h · 90d | 15m · 365d | custom retention |
| Support response | community | 4 hours | 1 hour · 24×7 | named engineer · 15 min |
| Uptime SLA | — | 99.9 % | 99.95 % | custom · BYOK |
Mumbai is live. More regions rolling out.
Your instance is provisioned into one region and stays there. Asia Pacific (Mumbai) is open today; additional regions come online as we bring AI compute up in each. Cross-region replication is a paid add-on, never a default — data residency is a yes-or-no knob. Regions outside the primary carry a 15 % uplift on compute + storage to cover their cost.
Common questions about pricing.
How is billing calculated?
One flat monthly compute fee (Whisper, Thunder, Storm, or Enterprise) plus a flat storage fee. Each tier includes a monthly quota of DB calls, natural-language queries, and outbound network. Above the quota you pay $0.0001/DB call, $0.02/natural-language query, and $0.12/GB outbound. Quotas reset on the 1st of each month; usage is metered to the second and billed monthly in arrears.
What does the 7-day trial include?
A Whisper-tier instance ($99/mo equivalent) with 20 GB of storage in the region you choose — enough headroom for one developer to evaluate the product end to end. Card is captured at instance create so the subscription converts seamlessly; you are not charged during the trial and can cancel any time before it ends. Trial is offered once per account on Whisper only — Thunder, Storm, and Enterprise tiers charge from day 1.
Can I add or remove add-ons mid-cycle?
Yes. Toggle any add-on on or off from the console at any time. Billing prorates to the day — enable mid-cycle and you only pay for the remaining days; disable mid-cycle and the line item is credited back. Usage-bound add-ons (Vector Search topk) meter only the calls you actually make.
Can I resize compute or storage later?
Yes. Vertical compute resize is a thirty-second task in the console — one reconnect, no data migration. Storage resize is live with no reconnect. Downgrades are allowed once per billing cycle.
Do you offer annual discounts?
Annual commit is 15 % off the equivalent monthly rate, paid in advance and credited monthly. Multi-year rates are available on request — email sales@originchain.ai.
What's in the Enterprise tier?
Negotiated quotas, customer-managed encryption keys (BYOK), a named support engineer, multi-region replication, custom retention windows, and a signed BAA, SOC 2, and DPA package. Pricing starts at the listed monthly minimum and scales with committed usage. Email sales@originchain.ai for custom terms.
Why is there a usage rate above the quota?
Two reasons. Natural-language queries cost real compute on third-party AI providers — a small markup keeps margins steady when a tenant's question mix gets harder. Outbound network past the included GB is metered straight from cloud cost rather than buried in the base price. Below the quota nothing extra is billed; above it, three line items are clearly itemised on the invoice and can be capped from the console.
What do I actually get on each tier?
A single-tenant OriginChain instance in your chosen region — its own HTTPS endpoint, its own bearer token, its own write-ahead log. We handle provisioning, TLS, rate limits, backups, metrics, and upgrades. You post sentences, get rows back.
Ninety seconds to an endpoint. No stack to wire up.
Pick a region, pick a tier, and we provision a single-tenant instance on AWS. The first query you send is the first query we'll show you how to write — in English.
- Sales sales@originchain.aivolume, SLA, procurement, BAA
- Support support@originchain.aitier-based response times