Pay for what you run — by the hour or the month
Buy compute nodes hourly as pay-as-you-go or reserve them monthly, then add usage-based AI, storage and egress. Transparent, itemised on one bill, and available in every OmniFabric region.
Available in every region
Spin up OmniFabric close to your users and your regulators. Your data is processed and stored only in the region you choose. List pricing is consistent across regions — talk to sales for committed-use and volume quotes.
Provisioning in Singapore — your data stays in-region
Buy compute by the hour or by the month
Sell-by-the-node compute that powers the whole fabric — data, models and agents. Burst with pay-as-you-go, or reserve monthly nodes and save.
Hourly nodes are billed per second with no commitment — ideal for bursts, batch jobs and dev. Monthly nodes are a flat reservation and run about 20% cheaper. Storage and data are decoupled from compute, so you can scale nodes up or down at any time without losing state.
Only pay for what you use
On top of compute, three metered services — each billed purely on consumption and itemised on the same invoice.
AI services
Route to 100+ hosted models — chat, embeddings and rerankers — through one API. Billed per million tokens; bring your own provider keys to pass through at cost.
Storage
One unified store for relational, vector and multimodal data. Compression and tiering are included, and reading your own data is always free.
Egress traffic
Outbound data transfer leaving your region for the public internet. Inbound and in-region traffic is always free.
Every node ships with the full Agent · Model · Data fabric — native vector search, HTAP queries, governance, lineage and SSO. Nothing is gated behind a higher tier, and all usage is metered to auditable records.
All prices in USD and exclude local taxes. Hourly usage is billed per second; monthly nodes are a fixed reservation.
Pricing questions, answered
Use hourly pay-as-you-go for spiky, short-lived or experimental workloads — you're billed per second with no commitment. Reserve monthly nodes for steady production and save around 20%. You can mix both in the same project and switch a node between the two at any time.
Every model call is metered by input and output tokens and priced per million. Use OmniFabric's routed models, or bring your own provider keys — in which case tokens pass through at the provider's cost with no markup, while still wrapped in the same lineage and governance.
List pricing is consistent across all seven regions. Your data is processed and stored only in the region you choose. For committed or high-volume use in a specific region, talk to sales for a tailored quote.
Yes — create a workspace for free, connect a source and ship a first agent. You only start paying when you provision a compute node or consume metered AI, storage or egress.
Storage is the data you keep in the fabric — tables, vectors and files — billed per GB-month. Egress is outbound traffic leaving the region for the public internet. Reads of your own data and traffic within a region are free.
Start free, scale when you're ready
Spin up a workspace, provision your first node by the hour, and only commit monthly once you're in production.