Regions · data handling
Know where a request can go
Choose Global for the lowest price or EU for the strictest data protection. Today both tiers run on the same EU GPU capacity in Finland; Global has lower priority. Global is moving onto a decentralized compute network that is rolling out and carries no customer traffic today. A key has an allowed tier and default region. An explicit S1-Region header may override the default only when the key permits it.
GLOBALMost affordable
Global — built to scale worldwide
The most affordable way to run decision models at scale. Global is moving onto a decentralized compute network: independent GPU operators worldwide, encrypted network traffic, every request and response cryptographically signed, and operator registration and performance scoring secured by blockchain technology.
Today
While the network rolls out, Global runs on the same EU GPU capacity as the EU tier — encrypted in transit, at lower priority than EU traffic. We will announce before any Global traffic moves to the network.
- s1-fast $0.025 · s1-pro $0.034 per 1M input tokens
- Output tokens free
- Best for high volume and non-sensitive data
Global pricing →EU
Highest data protection EU — for the strictest data-protection requirements
For teams with the highest data-protection demands or strict EU regulatory requirements. EU requests run only on contracted operators inside the EU, and inference data never leaves the EU.
- s1-fast $0.034 · s1-pro $0.040 per 1M input tokens
- Inference data stays in the EU (Finland)
- Contracted EU operators only — never the decentralized network
- Data Processing Agreement at signup, published sub-processor list
- Prompts and outputs are never stored or used for training
- Built for GDPR
EU site →Current routingToday both tiers run on the same EU GPU capacity in Finland. Global is lower priority and may briefly occupy at most one contracted base slot only while no EU request is waiting; at least one slot remains reserved for EU traffic. Global-only rejections do not trigger additional capacity purchases. Requests may otherwise be queued or rejected with a retryable error. We will announce before any Global traffic moves to the decentralized network; EU traffic never does. The published DPA covers both tiers.
Retry semantics
A retry does not replay a stored answer
For 24 hours, the record may retain only a keyed request digest, request ID, status and usage receipt; it does not store the request or answer. While an attempt is running, the same key and body return 202 in_progress. After a settled or settlement-uncertain attempt, the same key returns 409 idempotency_replay_unavailable with its receipt and no second charge. If an attempt is confirmed to have produced no answer and no charge, its claim is released and the same key and body may retry. Settlement uncertainty stays held until reconciled; recovery resolves stranded claims before expiry. A same key with a different body returns 409 idempotency_conflict.
Account controls
Usage history is not inference payload retention
The dashboard retention setting controls usage-history visibility. It does not control, extend or change inference-payload retention: EU payloads are not stored by default. Financial ledger records remain auditable account records.
See the trust overview and the privacy policy.
Processors and their locations are listed on the sub-processor page.