Utopai Launches PAI Production Platform and Utopai X Video Model, Linking Script Breakdowns to Generated Shots
VP Land
Oct 26, 2026 · 3 min read
Utopai Studios has launched PAI, a production intelligence platform for film and television that breaks a screenplay into scenes and shots, alongside Utopai X, a new video generation model that creates footage inside it. As Animation World Network reported, characters, locations and visual references stay tied to the project while footage is generated, and selected takes move into standard editing tools.
Scripts become connected scenes and shots. PAI keeps a project's characters, locations, shot plans and revision history available while a shot is developed, according to Utopai's launch post.
AI agents work from that project context. Utopai's Production Assistants help filmmakers plan shots, develop ideas and generate new iterations, while the filmmakers decide what gets made, kept and revised.
Takes, versions and handoff to post. Filmmakers can compare takes in the same environment, keep earlier versions and move selected footage into Adobe Premiere and DaVinci Resolve.
A second-place leaderboard ranking. Utopai says Utopai X ranks second on the Artificial Analysis leaderboard for text-to-video with audio, with an Elo score of 1,150.
Already in use on an animated feature. The studio built both in-house and is using them on its own productions, including Mike Bender's animated feature The Most Serious Fart.
PAI breaks a screenplay into scenes and shots and keeps characters, locations and decisions tied to the project
Once a screenplay is broken into scenes and shots, PAI keeps its characters, locations, objects, visual references and production decisions connected to the project. Production Assistants draw on that context when they help filmmakers plan shots and generate new iterations.
Utopai Studios' PAI and Utopai X. Image: Animation World Network
Utopai's launch post says the script breakdown and production assets stay available while a shot is developed, so a filmmaker can compare a new take with earlier versions. The company says that continuity reduces the need to rebuild the brief each time the team generates or revises footage.
Utopai Studios spells PAI out as Production Assistive Intelligence on its Utopai X product page.
Zijian He, Utopai Studios' chief scientific officer, framed the platform around creative control. "Technology is most useful when it gives filmmakers more control over their ideas," He said. "Our goal is to give artists the ability to explore more possibilities, iterate faster and make creative decisions with the same level of intention they bring to every other part of production."
Utopai X is post-trained on MiniMax H3, generates video with audio and hands selects to Premiere and Resolve
Utopai X generates video with audio from text prompts inside PAI. According to the Utopai X product page, the model is post-trained on MiniMax H3, users access it by signing in to or creating a PAI account, and generation uses credits, with rates listed on a separate PAI pricing page.
Filmmakers can compare multiple takes in the same environment, keep earlier versions as the work develops and move selected footage into post-production tools such as Premiere and DaVinci Resolve.
On image quality, Utopai says the model:
Handles complex visual behaviors, including reflections and caustics
Holds spatial consistency as shots grow more complex
Those are company claims. The announcement does not list resolution or clip length.
Utopai claims a second-place leaderboard spot and refines the tools on its own productions
Utopai cites an Elo score of 1,150 for Utopai X, nine points behind the top model, and says it is the highest-ranked model on that leaderboard from a U.S.-based company. Its product page adds that Artificial Analysis' launch analysis ranked the model first in Physics and Audio Synchronization and second in Lighting & Materials and Camera Control, and notes that rankings change over time.
Artificial Analysis ranks models through blind human-preference testing, in which participants compare outputs without knowing which model made them.
Compute provider GMI Cloud says Utopai trains Utopai X on its GPUs and serves it from there inside PAI. "We use GMI Cloud to train and serve our new model," said Jie Yang, Utopai's co-founder and CTO, who founded the studio with Cecilia Shen in 2022.
Because the tools are built inside a working studio, Utopai's technology teams use production data and feedback from its filmmakers and artists to refine the models, agents and workflows. The Most Serious Fart, the animated feature from Mike Bender, is one of the productions already running on PAI and Utopai X.
Utopai Studios develops, finances and produces its own film and television projects, co-produces with filmmakers, studios and rights holders, and runs Utopai Studios APAC and Utopai GmbH in Germany.
Billy Corben's 2006 documentary "Cocaine Cowboys" is getting a 20th anniversary re-release with new AI-generated reenactments from the AI studio New Territory Media, depicting events its former cartel members and hitm…
Kling AI made Kling 4.0 fully available through a refreshed creation interface on kling.ai and through MCP and CLI access, according to its launch post on X. The MCP and CLI routes open Kling 4.0 to agent and command-…
Sanxingdui: Future Past, which its producers bill as China's first AI-generated theatrical film, opened in Chinese cinemas to weak ticket sales, as a post from ToonHive highlighted. AFP, citing Maoyan, reported more t…