> At a Glance
> – Lightricks-2 generates 4K, 20-second, 50 fps clips with native audio on local hardware.
> – Open-weight model debuts on HuggingFace and ComfyUI, no cloud required.
> – Built with Nvidia and shown at CES 2026.
> Why it matters: Studios keep IP in-house while slashing cloud costs and wait times.
Most AI video tools are locked-away cloud beasts, but Lightricks just let the unicorn out: a second-generation model that runs natively on RTX-powered devices.
What Lightricks-2 Brings
The partnership with Nvidia bakes hardware-level optimization into every frame.
- Up to 20 seconds at 50 fps with 4K output
- Native audio baked in
- Targets individual filmmakers and major studios
Running locally keeps footage off third-party servers-crucial for entertainment giants guarding intellectual property.
How Local Processing Changes the Game
Typical video generators lean on Google-scale data centers, pushing average render times to 1-2 minutes and raising privacy flags.
| Benefit | Cloud Standard | Lightricks-2 Local |
|---|---|---|
| Data control | Shared | User-owned |
| Typical wait | 1-2 min | Faster on RTX rigs |
| IP risk | Higher | Minimal |
Shorter turnaround plus tighter security could speed up professional pipelines and trim budgets.
Key Takeaways
- Lightricks-2 is open-weight, not fully open-source, giving devs ingredient-level insight.
- Local generation slashes cloud fees and keeps sensitive footage private.
- Availability starts now on HuggingFace and ComfyUI for Nvidia RTX owners.
Expect studios and indie creators to test-drive the model as demand for secure, high-resolution AI video keeps rising.
