Fireworks AI

Where does this provider put your AI data — and who can compel its disclosure?

ResidencyUnited States us
SovereigntyUnited States us
Endpointsapi.fireworks.ai
Packagesfireworks (Python), @ai-sdk/fireworks (npm)

Data practices

How does this provider treat the data you send it? Advisory statements about documented practices — not legal advice.

Trains on customer API data by defaultNo source — Terms of Service (last updated 2026-07-01) §3.6 'Zero Data Retention': 'We will not use your Content to train our own models or to improve the Service' (retrieved 2026-10-10)
Retention (API inputs/outputs)Zero data retention by default for inference on open models: prompt and generation data exist only in volatile memory for the duration of the request, absent explicit opt-in (e.g. FireOptimizer logging). Exception: the Response API with store=True (its default) retains stored conversation data for 30 days, then auto-deletes; store=False or the DELETE endpoint is immediate. Non-content metadata (token counts) is logged. source — 'Zero data retention': prompt/generation data 'exist only in volatile memory for the duration of the request' without opt-in; 'Response API data retention': with store=True 'all stored conversation data is automatically deleted after 30 days' (retrieved 2026-10-10)
Subprocessor listhttps://trust.fireworks.ai/subprocessors — Named in the Fireworks DPA §6.2 as the canonical list of Authorised Sub-processors, with 30 days' change notice via Trust Center subscription (www.fireworks.ai/dpa) (retrieved 2026-10-10)
Enterprise tierEnterprise admins can enforce the zero-retention policy org-wide (rejecting requests that would persist or log content); the standard DPA (with SCCs/UK/Swiss addenda) is incorporated into the Terms for all customers (ToS §2.4), not enterprise-only. source — 'Enterprise account admins can additionally turn on the Zero Data Retention policy to reject requests and jobs that would persist or log customer content' (retrieved 2026-10-10)
Entry last reviewed2026-10-10