The first entry is always the pin's source. Overall confidence is a weighted average of how firmly each reference supports the start and end times used above; a reference counts half as much for every 180 days older than the newest.
Estimated: the technical report was first submitted 15 December 2025 (arXiv v1) and Vercel lists the model on 16 December; other sources give 18 or 23 December, so the exact public launch day is not confirmed.
Explainer on the model's capabilities.[1]
Something missing or wrong? Say it in your own words: a link that backs this pin up, a different start or end date and why, or a fact it lacks or gets wrong. The AI checks it against this pin's sources, searches for better ones, and adds any page that backs you up. The pin's own sources still count most. A picture that shows something else, or shows it badly, is looked at too, and moved down or replaced.
| About USD 0.26 per 5 s 720p video with audio; USD 2.40 per million video tokens with audio at other resolutions |
Release, architecture and audio rows from the ByteDance report; resolution, frame rate, length and price from fal.ai's page and API schema. No generation-speed figure is published by ByteDance; fal's page states roughly 30-45 s for a 5 s clip, varying by hardware, which is left out as a host estimate. · source
| Spec | Value |
|---|---|
| fal.ai, 720p, 5 s, with audio | About USD 0.26 per video |
| fal.ai, other resolutions | USD 2.40 per 1 million video tokens with audio |
As published by fal.ai; read 24 October 2026. · source
Translations completed using Gemini 3.8 Flash.