Skip to main content
The Video Datalake is pay-as-you-go against a prepaid balance. Management and read-from-database operations are free — you pay only for the APIs that do real work: indexing, search, media generation, and storage. Check your balance with Get Balance. When it reaches zero, billable requests fail with 402 quota_exceeded. Failed requests are never billed.

​
Free APIs

Every collection / video / person management call and every list or detail read is free:

​
Priced APIs

Indexing is priced per minute of video and always embeds with OmniRetriever (3072-dim, multilingual). Everything else is per call, per image, or per GB.
Indexing cost is per minute at fps = 1.0 (the default) and scales roughly linearly with fps. It covers standard indexing (OmniRetriever embeddings, audio transcription, OmniCaptioner captions & summary). Collections that enable face recognition or safety detection run additional detection on every indexed video, at additional per-minute cost.
Search rerank: true runs an extra cross-encoder pass and is billed on top of the base search call.

​
Billing notes

  • Failed requests are not billed — the balance is only deducted on success.
  • Idempotent replays are not double-billed — resubmitting with the same Idempotency-Key (or idempotency_key on uploads) returns the first result at no extra charge.
  • Cancelled ingests — an ingest cancelled before the index phase starts is not billed for indexing.
  • Insufficient balance — billable requests fail immediately with 402 quota_exceeded; nothing is queued.
Stored data is deleted 30 days after your balance reaches zero. We send reminders on day 15 and day 29. Top up before day 30 and nothing is lost — after that, stored videos, frames, vectors, and face images are permanently deleted and have to be re-indexed.

​
Plans

Pay-as-you-go needs no plan. A $20 / month starter plan is also available — it bundles 10,000 free simple-API calls, 2 QPS, 1 project, and 2 API keys, with all indexing / search / AI processing billed at the per-API rates above. Response-size caps apply on every plan — see Errors & Rate Limits. For higher rate limits, more projects, volume discounts, or an SLA, contact sales.

​
Manage balance

View and top up your balance in the Memories.ai Console: https://console.memories.ai/stripe For enterprise pricing, contact support@memories.ai.