What changed: eleven kinds of change
Each page takes one kind of change and lists every entry of that kind across the lines, with pages for single lines where there are three or more. PixVerse's first-and-last-frame page stays at two entries, after the changelog was read again on 2026-10-25. As of 2026-09-25.
An entry can count under several kinds when its note describes several changes.
Kinds are assigned from the note's own words.
1In this section
- API, plans and availability — API, plans and availability in AI video release notes: 22 entries from 10 model lines, 21 of them dated September 2024 to August 2026.
- Camera and several shots — Camera and several shots in AI video release notes: 21 dated entries from 10 model lines, October 2024 to August 2026.
- Clip length and extension — Clip length and extension in AI video release notes: 38 entries from 14 model lines, 32 of them dated September 2024 to August 2026.
- Editing video that exists — Editing video that exists in AI video release notes: 25 entries from 11 model lines, 24 of them dated September 2024 to August 2026.
- Faster and lighter tiers — Faster and lighter tiers in AI video release notes: 33 entries from 10 model lines, 29 of them dated August 2024 to September 2026.
- Motion and performance — Motion and performance in AI video release notes: 16 dated entries from 9 model lines, July 2024 to June 2026.
- New models and snapshots — New models and snapshots in AI video release notes: 70 dated entries from 17 model lines, July 2024 to September 2026.
- References and consistency — References and consistency in AI video release notes: 34 entries from 13 model lines, 33 of them dated December 2024 to September 2026.
- Resolution, frame rate and shape — Resolution, frame rate and shape in AI video release notes: 40 entries from 14 model lines, 36 of them dated September 2024 to September 2026.
- Sound, voice and lip-sync — Sound, voice and lip-sync in AI video release notes: 22 dated entries from 7 model lines, December 2024 to August 2026.
- Start, end and middle frames — Start, end and middle frames in AI video release notes: 30 entries from 14 model lines, 29 of them dated July 2024 to August 2026.
2Using the models, not reading about them
Release notes such as “What changed” matter once a pipeline runs the models; SceneMixer's guides describe how it does:
Doubao Seed 2.1 Turbo is the tier every account uses for text work until you change it: it reads the script, writes with AI when you ask, and generates each episode’s storyboard, at 50% of the base text rate.
SceneMixer, Doubao Seed 2.1 Turbo: the default text tier (read 2026-10-13)
Recommended is the badge on the tier the site suggests for most character sheets; the default for new accounts stays the cheaper tier.
SceneMixer, Qwen-Image 3.0 Pro: the recommended tier — “Is Recommended the same as default?” (read 2026-10-13)