Mastodon alt text: 10,000 characters and a video transcript

Mastodon 4.6.0 changed the alt text limit for media attachments to 10,000 characters, which leaves room for a full transcript on a video. Import the clip to your workspace (POST /v1/media-imports), then run POST /v1/video-inspect on its media.sume.com URL with transcribe: true to get the spoken text, then edit it into a description yourself.

Mastodon facts are from its changelog and Sume facts from Video inspect, read 2026-09-30.

What changed in Mastodon 4.6.0?

The changelog entry dated 2026-06-17 lists three relevant lines: the alt text limit for media attachments is now 10,000 characters, the description limit for remote attachments went from 1500 to 10000, and duration was added to the ActivityPub representation of media attachments. Your instance's own limit may differ if it runs another version.

What does Sume give me to write from?

How do I turn that into alt text?

Transcribe, then add what a listener cannot hear: who is on screen, text that appears, and visual events. A transcript alone describes the audio, not the picture, so use the stills to write the visual part. Keep the final string under the limit by checking its length in your script. The transcription half is billed at Sume STT's rate listed in the docs (confirm live in GET /v1/catalog), while probe and stills are unbilled.

What if the clip is silent?

A silent clip with transcribe: true fails with inspect_source_has_no_audio. Check probe.has_audio first with a frames: false inspect. For a silent video, write the description from the stills. Posting limits for video itself are covered in Mastodon video size limit.