Focusing on spatial intelligence! "AI Godmother" Fei-Fei Li launches her first commercial world model

Focusing on spatial intelligence! "AI Godmother" Fei-Fei Li launches her first commercial world model

"AI Godmother" Fei-Fei Li’s startup has taken a key step towards commercialization in the "world model" track. Her company, World Labs, has officially launched its first commercial product, Marble, marking the expansion of artificial intelligence from abstract language and text comprehension to spatial intelligence that simulates and interacts with the physical world, with the intention to gain a lead in the next generation AI technology race.

On November 12, World Labs, co-founded by Stanford Professor Fei-Fei Li, announced the public launch of Marble, its first product driven by a multimodal world model. This model can utilize text, photos, videos, or 3D layouts as inputs to generate editable and downloadable 3D interactive environments, giving World Labs an early edge in competing with tech giants like Google.

Compared to the limited preview version released two months ago, the official Marble version has achieved significant functional upgrades, including support for larger-scale multimodal inputs and the launch of a creative hub called Marble Labs. Business-wise, it adopts a freemium and paid subscription model, aiming for rapid market entry and user base growth.

The launch of Marble directly targets industries with huge demand for 3D content, such as game development, visual effects (VFX), and virtual reality (VR). It offers creators a brand-new asset generation tool while also sparking in-depth industry discussions on how AI might reshape creative workflows and its potential impact on jobs.

First to Commercialize, Building Differentiation Amid Competition

In the burgeoning world model field, Marble’s commercial release helps it stand out from numerous research projects and technology demos.

Currently, models like Google’s Genie are still in limited research preview, and products from startups Decart and Odyssey remain at free demo versions. According to Bloomberg, Marble is the first product in this track to be commercially applied. Its core differentiation lies in its ability to generate persistent, downloadable 3D environments, in contrast to World Labs’ own real-time model RTFM (Real-Time Foundation Model), which dynamically generates worlds as users explore.

The company states that this design significantly reduces scene deformation and inconsistency, while allowing users to export results in professional formats like Gaussian splats, meshes, or videos, to integrate into existing workflows.

Marble's input and output pipelineImage source: World Labs

Focus on Creative Control and Multimodal Input

To address the pain point of “uncontrollable” AI-generated content, Marble puts “creative control” at the core of its product. World Labs co-founder Justin Johnson told TechCrunch, “We don’t want machines to take over completely and deprive creators of their creativity.”

To this end, Marble offers high flexibility on both input and editing sides.

Input side: The official version supports users uploading multiple images or short videos from different angles to present a space, leading to more realistic digital twins. This solves the limitation of the preview version, which only supported single-image input and required the model to “imagine” vast amounts of detail.Editing side: The product launches an experimental 3D editor called “Chisel.” This feature allows users to first build a rough spatial structure (like walls and blocks) and then fill in visual styles with text prompts, achieving a separation of structure and style. Johnson explained, “I can enter the scene and directly grab the 3D block representing a sofa and move it.”

Additionally, users can “expand” already-generated worlds or splice multiple independent worlds into a larger space under “composer mode,” further enhancing creative freedom.

Spaceship environment created with Marble, with overlaid text prompt (Note how the lighting realistically reflects on the hub walls).
Image source: World Labs/TechCrunch

Clear Business Model Targeting Three Major Application Scenarios

Marble has a clear business path, offering tiered subscriptions to meet different user needs. Its subscription plans come in four tiers: The free version provides four generations per month; Standard ($20/month) and Pro ($35/month) offer more generations and advanced features; and the Flagship ($95/month) unlocks all features, with 75 generations and commercial usage rights.

In the short term, World Labs is targeting three main markets:

Game development: Johnson believes developers can use Marble to generate background environments and mood spaces, then import these assets into game engines like Unity or Unreal for further development—supplementing, rather than replacing, existing pipelines.Visual Effects: Compared to AI video generators, Marble’s 3D assets provide artists with precise scene control and camera work, avoiding image inconsistency issues.Virtual Reality (VR): Though VR isn’t the current top priority, Johnson noted the industry has an “extreme hunger” for high-quality content. All worlds generated by Marble are already compatible with mainstream headset devices like Vision Pro and Quest 3.

However, market acceptance of this technology isn’t wholly optimistic. According to a recent Game Developers Conference (GDC) survey, one third of respondents believe generative AI negatively impacts the gaming industry, with concerns mainly around intellectual property and declining content quality.

Fei-Fei Li’s Ultimate Vision: Towards "Spatial Intelligence"

Behind Marble’s release is Fei-Fei Li’s grand vision for the future of AI—achieving “Spatial Intelligence.”

In a recent article, Fei-Fei Li outlined that while current large language models excel at handling abstract knowledge, they lack genuine understanding of the physical world, akin to “walking in darkness.” She believes endowing machines with the ability to understand, navigate, and interact with the 3D world—spatial intelligence—is key to achieving truly general AI. World models are the core technology for reaching this goal.

According to her vision, spatially intelligent world models will transform multiple industries in stages:

Short-term: Empower creative industries, giving filmmakers, gamers, and architects powerful tools.Mid-term: Advance embodied intelligent robots, training them at scale in simulated environments to become collaborative partners for humans.Long-term: Spark revolutions in science, medicine, and education, enhancing human experts’ capabilities through simulated experiments and diagnostic assistance.

“Without spatial intelligence, our dream of truly intelligent machines will remain incomplete,” Li wrote. This vision is strongly supported by capital markets. Since its founding in 2024, World Labs has quickly raised about $230 million in funding and reached a valuation of $1 billion, becoming a “unicorn.” Its investors include a16z, Nvidia NVentures, AMD Ventures, Intel Capital, and other industry giants and top VC firms.

Risk Warning and DisclaimerThe market carries risks; investment should be cautious. This article does not constitute personal investment advice, nor does it consider the specific investment objectives, financial status, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their own circumstances. Investing accordingly is at your own risk.