Skip to main content
This page collects operational details that apply across the Engramme API.

​
Rate Limits

Requests are subject to gateway and application-level limits. The public gateway currently enforces: If you receive 429 Too Many Requests, retry with exponential backoff. Response bodies can vary by which layer returns the limit. Application-level limits may return:

​
Request Limits

item_id may contain word characters plus -, _, =, !, and :.

​
Status Codes

​
Training Conflicts

If you provide your own item_id, /v1/memorize may return 409 when the item is already being processed. Use generated IDs for simple training requests. Use stable unique item_id values when you want deduplication or idempotent behavior.

​
Async Training

/v1/memorize returns after the file is accepted and asynchronous training has started. Memories may not be immediately available through /v1/recall. If recall returns no results right after a training request, retry after a short delay.