bestofnow← All categories

Video Generation

Ranked by momentum — measured, not bought.

  1. 1
    Google Veo↗ Rising

    Google's video model with the strongest realism and the only native 48kHz synchronized dialogue, in Lite, Fast, and Quality tiers.

    Score 71.67

    Why it's rising

    Google Veo's engagement has increased 158% over the past week.

  2. 2
    Kling↗ Rising

    Kuaishou's video model for cinematic motion and multi-shot scenes, with native 4K, Motion Control, and audio synced across cuts.

    Score 63.64

    Why it's rising

    Kling's engagement has increased 161% over the past seven days.

  3. 3
    Synthesia↘ Cooling

    Avatar video platform for corporate training and localization, turning scripts into presenter-led videos with multilingual lip-sync.

    Score 37.19

    Why it's cooling

    Synthesia's engagement has declined 59% over the past week, with recent Hacker News discussion showing 15 stories, 169 points, and 81 comments.

  4. 4
    SeedanceWatchlist — not enough signal yet

    ByteDance's video generation model that produces clips with synchronized audio in a single pass, known for balance, speed, and strong physics.

    Score 33.45

    What the data shows

    Seedance has generated 6 Hacker News stories with 34 points and 10 comments over the past 7 days, but lacks sufficient history to establish a reliable momentum trend.

  5. 5
    WanWatchlist — not enough signal yet

    Alibaba's open-source video model, free to self-host, valued for fast iteration and unrestricted prompts at low cost.

    Score 27.00

    What the data shows

    Wan has 16,658 stars and 3,001 forks on GitHub with a most recent push on March 5, 2026.

  6. 6
    HappyHorseWatchlist — not enough signal yet

    Alibaba's video generation model topping the quality leaderboards, with joint audio-video output and seven-language lip-sync.

    Score 8.54

    What the data shows

    HappyHorse has received minimal engagement with one story and one point on Hacker News over the past seven days.

Watchlist — building baseline

Not enough cross-source signal yet for a score or a trend. These are measured nightly — they move up the moment the data does.

  • Higgsfield

    Creative suite that runs 15+ video models like Veo, Kling, and Seedance under one subscription, adding character consistency and camera control.

  • Runway

    Pro video-creation suite with its own Gen models, known for the best directorial control: motion brushes, camera moves, and scene consistency.

  • HeyGen

    Avatar video app for marketing and localization, generating talking-head videos and translations from a script in 175+ languages.

  • Luma Dream Machine

    Luma's video app built on its Ray3 model, noted for elegant, atmospheric output and being first with native 16-bit HDR.