> ## Documentation Index
> Fetch the complete documentation index at: https://docs.eversince.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Limits

> Every ceiling a caller meets, from call timing and request size to file sizes, counts and rates.

|                                |                                                                                                                                                             |
| ------------------------------ | ----------------------------------------------------------------------------------------------------------------------------------------------------------- |
| A tool call                    | answers within about 30 seconds or returns a job; the API allows up to 300 seconds                                                                          |
| `get_jobs` long-poll           | `wait_seconds` up to 25, inside the 30 seconds some clients allow a call                                                                                    |
| Request body                   | 9 MB; `import_media` in `file` mode carries up to 6 MB decoded                                                                                              |
| Uploads                        | image 20 MB, audio 500 MB, video 10 GB, document 25 MB; a presigned upload carries up to 5 GB, and a larger file goes through `import_media` in `link` mode |
| Import by URL, tool or route   | image 20 MB, video 200 MB, audio 50 MB, document 25 MB; 20 URLs per call                                                                                    |
| Transcription within the call  | media under five minutes answers within the call; longer returns a job                                                                                      |
| `inspect_media`                | 12 refs or URLs per call                                                                                                                                    |
| `get_items` and `update_items` | 25 items per call                                                                                                                                           |
| `delete_items`                 | 100 refs per call; a longer list is refused whole                                                                                                           |
| `show_media`                   | 10 refs per call                                                                                                                                            |
| A document read                | about 60,000 characters of clips or layers, then `next_cursor`                                                                                              |
| A skill body                   | 25,000 tokens                                                                                                                                               |
| Revisions kept                 | 50 on a timeline, 20 on a canvas                                                                                                                            |
| API keys                       | 10 per account                                                                                                                                              |
| Webhook endpoints              | 10 per workspace; 10 seconds per delivery; 8 attempts over about a day                                                                                      |
| Rate                           | 1,000 requests a minute per account across every interface; `Retry-After` on a 429                                                                          |
| Upscale                        | video up to five minutes at a time                                                                                                                          |

The file size ceilings are also returned live by `GET /api/v1/tools` in `ingest_limits`.