Skip to content

AI videos

Type an idea (or pick an image you already have) and get back a short video clip you can use in a post.

Why use it

You don't need any video editing skills or extra software. Describe what you want in plain words, and Sosyabot makes a short clip for you in a few minutes.

Step by step

Describe or choose your source

Write a short sentence about the video you want, like "a cup of coffee steaming on a wooden table." If you'd rather start from a picture, choose one of your images to bring it to life.

Start the video

Click the button to start. The video begins generating right away.

Wait a few minutes

Making a video takes a little time. You can keep working elsewhere in Sosyabot and come back. The clip will be ready shortly.

It attaches to your post

When it's done, the finished clip lands in your library and attaches to your post, ready to share.

WARNING

The 20 credits are charged the moment you start, before the video is made. If something goes wrong and the video fails, the credits are still used. Also expect to wait a few minutes for each clip.

TIP

Keep your description short and clear. One simple scene works best, like "ocean waves at sunset."

Good to know

  • A video costs 20 credits each time you start one.
  • For comparison, a quick text answer costs 1 credit and an image costs 5 credits.
  • You're charged when you start, even if the video later fails.
  • Each video takes a few minutes to finish.
For developers

Video generation is asynchronous. POST /api/v1/ai/videos submits a job; a BullMQ ai-video worker polls the configured provider (fal.ai default model fal-ai/veo3/fast, or Runway gen3a_turbo) every 5 seconds until completed or failed, with a 10-minute per-job timeout. List jobs with GET /api/v1/ai/videos and poll one with GET /api/v1/ai/videos/:id. Status flow: pending → processing → completed | failed. The 20-credit charge happens at submit time, not on completion. Request body accepts prompt, optional provider, duration (1–60s, default 5), aspect_ratio, and model.