World Labs, the startup founded by AI pioneer Fei-Fei Li, is launching its first business world model product. Marble is now available via freemium and paid tiers that allow users turn text prompts, photos, videos, 3D layouts, or panoramas into editable, downloadable 3D environments.
The launch of the generative world model, first released in limited beta preview two months ago, comes just a little over a 12 months after World Labs got here out of stealth with $230 million in funding, and puts the startup ahead of competitors constructing world models. World models are AI systems that generate an internal representation of an environment, and will be used to predict future outcomes and plan actions.
Startups like Decart and Odyssey have released free demos, and Google’s Genie continues to be in limited research preview. Marble differs from these — and even World Labs’ own real-time model, RTFM — since it creates persistent, downloadable 3D environments fairly than generating worlds on-the-fly as you explore. This, the corporate says, leads to less morphing or inconsistency, and lets users export worlds as Gaussian splats, meshes, or videos.
Marble can also be the primary model of its kind to supply AI-native editing tools and a hybrid 3D editor that lets users block out spatial structures before AI fills within the visual details.
“It is a brand latest category of model that’s generating 3D worlds, and that is something that’s going to get well over time. It’s something we’ve already improved quite loads,” Justin Johnson, co-founder of World Labs, told TechCrunch.
Last December, World Labs showed how its early models could generate interactive 3D scenes based on a single image. While impressive, the somewhat cartoonish scenes weren’t fully explorable since movements were limited to a small area, and there have been occasional rendering errors.
In my trial of the beta preview, I discovered Marble generated impressive worlds from image prompts alone — from game-like environments to photorealistic versions of my front room. Scenes morphed at the sides, though that’s apparently been improved in today’s launch. That said, a world I’d generated within the beta using a single prompt looked higher and matched my intent more closely than the identical prompt does now.
Techcrunch event
San Francisco
|
October 13-15, 2026
I haven’t yet tested the editing features, though Johnson says they make Marble practical for near-term gaming, VFX, and virtual reality (VR) projects.
“One in all our predominant themes for Marble going forward is creative control,” Johnson said. “There should at all times be a fast pathway to generate something, but it is best to find a way to dive even deeper and get plenty of control over the things that you simply’re generating. You don’t want the machine to simply take the wheel and pull all that creativity away from you.”
Marble’s tackle creative control starts with input flexibility. The beta only accepted single images, forcing the model to invent unseen details for a 360-degree view. With the total launch, users can now upload multiple images or short clips to indicate an area from different angles and have the model generate fairly realistic digital twins.
Then we’ve Chisel, an experimental 3D editor that lets users block out coarse spatial layouts (think partitions, boxes, or planes) after which add text prompts to guide the visual style. Marble generates the world, decoupling structure from style — just like how HTML provides the structure of an internet site and CSS adds in color. Unlike text-based editing, Chisel permits you to directly manipulate objects.
“I can just go in there and grab the 3D block that represents the couch and move it some other place,” Johnson said.
One other latest feature that provides you more editing control is the power to expand a world.
“When you generate a world, you’ll be able to expand it as much as once,” Johnson said. “If you move to a bit of the world that’s beginning to break apart, you’ll be able to mainly tell the model to expand there or generate more world within the vicinity of where you currently are, after which it may possibly add more detail in that region.”
Users who need to create extremely large spaces can mix multiple worlds with “composer mode.” Johnson demonstrated this for me with two worlds he had already built — a room made from cheese with grape chairs, and one other of a futuristic meeting room in space.
The trail to spatial intelligence
Marble is out there via 4 subscription tiers: Free (4 generations from text, image, or panorama), Standard ($20/month, 12 generations plus multi-image/video input and advanced editing), Pro ($35/month, 25 generations with scene expansion and business rights), and Max ($95/month, all features and 75 generations).
Johnson thinks the initial use cases for Marble will probably be gaming, visual effects for film, and virtual reality.
Game developers have mixed feelings in regards to the tech. A recent Game Developers Conference survey found a 3rd of respondents believed generative AI has a negative impact on the games industry — 12% greater than the survey indicated 12 months earlier. Mental property theft, energy consumption, and a decrease in quality from AI-generated content were among the many top concerns aired. And last 12 months, a Wired investigation found game studios like Activision Blizzard are using AI to chop corners and combat attrition.
In gaming, Johnson sees developers using Marble to generate background environments and ambient spaces after which importing those assets into game engines like Unity or Unreal Engine so as to add interactive elements, logic, and code.
“It’s not designed to switch your complete existing pipeline for gaming, but to simply offer you assets that you could drop into that pipeline,” he said.
For VFX work, Marble sidesteps the inconsistency and poor camera control that plague AI video generators, per Johnson. Its 3D assets let artists stage scenes and control camera movements with frame-perfect precision, he said.
While Johnson said World Labs isn’t specializing in virtual reality (VR) applications at once, he noted the industry is “starved for content” and excited in regards to the launch. Marble is already compatible with the Vision Pro and Quest 3 VR headsets, and each generated world will be viewed in VR today.
Marble may additionally have potential use cases for robotics. Johnson noted that unlike image and video generation, robotics doesn’t take pleasure in a big repository of coaching data. But with generators like Marble, it becomes easier to simulate training environments.
In accordance with a recent manifesto by Fei-Fei Li, CEO and co-founder of World Labs, Marble represents step one toward creating “a very spatially intelligent world model.”
Li believes “the following generation of world models will enable machines to realize spatial intelligence on a completely latest level.” If large language models can teach machines to read and write, Li hopes systems like Marble can teach them to see and construct. She says the power to grasp how things exist and interact in three-dimensional spaces can eventually help machines make breakthroughs beyond gaming and robotics, and even into science and medicine.
“Our dreams of truly intelligent machines won’t be complete without spatial intelligence,” Li wrote.
Got a sensitive tip or confidential documents? We’re reporting on the inner workings of the AI industry — from the businesses shaping its future to the people impacted by their decisions. Reach out to Rebecca Bellan at rebecca.bellan@techcrunch.com or Russell Brandom at russell.brandom@techcrunch.com. For secure communication, you’ll be able to contact them via Signal at @rebeccabellan.491 and russellbrandom.49.