Fei-Fei Li, renowned as the "AI Godmother," has taken a pivotal step toward commercializing her startup’s "world model" technology. Her company, World Labs, unveiled its first commercial product, Marble, marking a shift in artificial intelligence from abstract text comprehension to spatial intelligence—simulating and interacting with the physical world—to gain an edge in next-gen AI competition.

On November 12, World Labs, co-founded by Stanford professor Fei-Fei Li, announced the public release of Marble, a multimodal world model-powered product. The model processes diverse inputs like text, photos, videos, or 3D layouts to generate editable and downloadable 3D interactive environments, positioning World Labs ahead of tech giants like Google in this emerging field.

Compared to its limited preview two months ago, the official Marble version now supports larger-scale multimodal inputs and introduces Marble Labs, a creative hub. The product adopts a freemium and subscription-based model to accelerate market penetration and user adoption.

Marble targets industries with high demand for 3D content, such as game development, visual effects (VFX), and virtual reality (VR). While it offers creators a novel asset-generation tool, it also sparks debates about AI’s potential to reshape creative workflows and its impact on jobs.

**Pioneering Commercialization to Stand Out** In the nascent world-model sector, Marble’s commercial launch distinguishes it from research projects and tech demos. Competitors like Google’s Genie remain in limited research previews, while startups like Decart and Odyssey offer only free demos. Marble is reportedly the first commercially deployed product in this space, with its key differentiation being persistent, downloadable 3D environments—unlike World Labs’ real-time model RTFM (Real-Time Foundation Model), which generates worlds dynamically.

The company claims this design minimizes scene distortions and inconsistencies, allowing exports in professional formats like Gaussian splats, meshes, or videos for seamless workflow integration.

**Creative Control and Multimodal Inputs** Addressing AI’s "uncontrollable" output, Marble prioritizes creator autonomy. Co-founder Justin Johnson told TechCrunch, "We don’t want machines to take over creativity."

Input Flexibility: The official version supports multi-image or short-video uploads to create realistic digital twins, overcoming the preview’s single-image limitation. Editing Tools: The experimental 3D editor "Chisel" lets users draft spatial structures (e.g., walls, blocks) and refine styles via text prompts. Johnson explained, "You can grab a 3D block representing a sofa and move it directly."

Users can also expand generated worlds or merge them in "Composer Mode" for greater creative freedom.

**Clear Monetization Strategy for Three Key Sectors** Marble offers tiered subscriptions: - Free: 4 monthly generations - Standard ($20/month) and Pro ($35/month): More generations and advanced features - Flagship ($95/month): Full access, 75 generations, and commercial rights.

Initial focus areas: 1. **Game Dev**: Marble-generated assets can supplement pipelines in engines like Unity or Unreal. 2. **VFX**: Provides precise scene control, avoiding inconsistencies in AI-generated videos. 3. **VR**: Though not a priority, Marble’s outputs are compatible with Vision Pro and Quest 3.

However, skepticism persists. A recent GDC survey found 33% of respondents view generative AI negatively, citing IP and quality concerns.

**Fei-Fei Li’s Vision: The Path to Spatial Intelligence** Behind Marble lies Li’s broader ambition: advancing "spatial intelligence." In a recent article, she argued that while large language models excel in abstract knowledge, they lack physical-world understanding. Spatial intelligence—enabling machines to navigate and interact with 3D environments—is critical for true AGI.

Her roadmap for world models: - **Short-term**: Empower creative industries (film, gaming, architecture). - **Mid-term**: Advance embodied AI robots via simulated training. - **Long-term**: Revolutionize science, healthcare, and education through simulations.

"Without spatial intelligence, our dream of truly intelligent machines is incomplete," Li wrote. The vision has attracted robust funding—World Labs, founded in 2024, has raised $230M at a $1B+ valuation, backed by a16z, NVentures, AMD Ventures, and Intel Capital.