Generative video models

Models that generate video, and now synchronized audio, from text or images; moving toward interactive 'world simulators'.

AI & roboticsSoftware
  1. 2022

    First version

    Meta (Make-A-Video), Google (Imagen Video, Veo), Runway, OpenAI (Sora)

    Image: Meta Platforms · Public domain · Wikimedia Commons
  2. 2023

    Runway Gen-2

    Text-to-video leaves the lab: anyone can type a prompt in a browser or phone app and get a short clip.

    Image: Juhun87621 · CC0 · Wikimedia Commons
  3. 2024

    Sora preview, Veo

    Minute-long coherent clips framed as world simulation.

    Image: OpenAI · © OpenAI; unclear reuse licence · Wayback Machine: og:image of sora.com, captured…
  4. 2025

    Veo 3

    Generates synchronized sound, dialogue and effects together with the video, ending silent clips.

    Image: Google DeepMind · © Google LLC; trademark, no licence stated · Wayback Machine capture (31 May 2025) of deepmind.google/models/veo
  5. 2025
    Genie 3 (interactive world model)

    Genie 3 (interactive world model)

    Generates explorable 3D worlds in real time at 720p that users can navigate and alter.

    Image: Google DeepMind · © Google DeepMind; no licence stated · Google DeepMind Genie 3 announcement video frame (5 Aug 2025)

See it on the timeline