wired.com web signal

Rogue 1.0 Wraps Sora, Kling, Google as SFW Backends

TL;DR

  • Rogue 1.0 accepts up to 13 character references per scene and outputs 1080p video with native audio.
  • The platform's frame rate is adjustable between 24 and 60 fps, and clips extend to 60 seconds.
  • Safe-for-work jobs route to Sora, Kling, Luma, Google, MiniMax, and Seeddance; NSFW work is reserved for the in-house Rogue model.

The uncensored AI video model Rogue 1.0 accepts up to 13 character references per scene and outputs 1080p video with native audio, according to Rogue Studio's product page. Wired profiled the studio behind it as part of a small wave of AI video companies positioning themselves outside the content guardrails of the major labs.

The product page lists frame rates adjustable between 24 and 60 fps and clip extensions up to 60 seconds. The company, Fangenie LLC, is based in Beverly Hills, and the wide release was July 1, 2026. Inside the platform, users can route safe-for-work jobs to Sora, Kling, Luma, Google, MiniMax, and Seeddance, with Rogue's own model reserved for the NSFW work the other providers refuse. The site describes Rogue as "the world's most powerful uncensored video model."

This is the third AI video story on our tracker in as many days, landing just after Aronofsky and Blomkamp built new studios on top of US and Chinese base models. Rogue is the same market from the opposite end: instead of prestige auteurs licensing frontier tools, a Beverly Hills LLC pitching creative-freedom software to buyers the frontier labs will not serve.