Decart Launches Oasis 3 to Advance Real-Time World Modeling
AI startup Decart has introduced Oasis 3, an interactive world model capable of generating photorealistic, real-time driving environments. By providing API access to developers, the company aims to establish a robust ecosystem for physical AI, specifically targeting autonomous vehicle manufacturers who require scalable, diverse simulations of rare driving scenarios. This release marks a strategic expansion from the company’s previous work in e-commerce and live streaming, leveraging its existing foundation model, Lucy.
What sets Oasis 3 apart in a crowded market—which includes competitors like Google’s Genie 3 and World Labs’ Marble—is its focus on computational efficiency. Decart utilizes its proprietary Decart Optimization Stack (DOS) to streamline performance across major hardware providers, including Nvidia, Amazon, and Google. This vertical integration allows the company to offer significantly lower operational costs than its rivals, a factor that likely contributed to its recent $300 million funding round and $4 billion valuation.
The implications of this technology are significant for the future of robotics and autonomous systems. By enabling infinite, interactive scenario generation, Oasis 3 allows developers to stress-test AI agents against countless edge cases that are difficult or dangerous to replicate in the real world. However, the technology is not without its limitations; early testing indicates that while the initial scene generation is highly impressive, the thematic and visual integrity of the environment can degrade over extended periods of interaction.
Ultimately, Decart is positioning itself as a critical infrastructure provider for the next generation of physical AI. With strategic backing from industry giants like Toyota, Adobe, and Nvidia, the company is betting that its combination of high-fidelity realism and cost-effective scalability will make it the preferred platform for developers building the virtual worlds of tomorrow.