Odyssey is pioneering general-purpose world models with Crusoe’s powerful, scalable AI cloud

Odyssey is an AI lab pioneering general-purpose world models; a new form of multimodal intelligence unlocking entirely new consumer, enterprise, and intelligence applications. World models are a major frontier in AI, and Odyssey is leading the way with breakthrough models like Odyssey-2 Pro.

Simple, causal prediction objectives have led to surprisingly general intelligence. We’re now seeing this approach extend beyond language to world models, and the emergence of world simulators. A world simulator (like Odyssey-2 Pro) is a model trained to predict how the world evolves over time, frame-by-frame, using large amounts of video and interaction data. These models output continuous, interactive audio-visual simulations that will unlock entirely new consumer, enterprise, and intelligence applications.

The challenge: Seeking a foundational partner for a frontier mission

As a startup building a product where the AI model is the entire company, Odyssey’s choice of a cloud infrastructure provider was a foundational business decision. Their evaluation was not about solving a single technical problem but about finding a long-term, strategic partner. The key criteria were clear: they needed exceptional performance, cost-effectiveness, reliability, and  commercial flexibility.

Initial conversations with traditional hyperscaler cloud providers proved frustrating. Odyssey encountered rigid, multi-year contracts and a lack of transparency that felt misaligned with the agile, fast-evolving needs of an AI innovator.

They needed a provider who could offer a seamless infrastructure roadmap, evolving with them from NVIDIA H100s to H200s and onward to next-generation architectures like the GB200. What they wanted was a partner that combined the agility of a startup with the robust infrastructure of a major player.

The solution: A bespoke partnership on Crusoe's energy-first AI cloud

Odyssey found exactly what they were looking for with Crusoe. The relationship was immediately different, built on a foundation of deep collaboration and flexibility. Odyssey describes Crusoe as "an extension of their team," a curated relationship involving joint engineering and product development efforts. This began with an agreement that allowed Odyssey to scale its commitments as its needs grew — a stark contrast to the rigid proposals from competitors.

On this foundation, Odyssey built its end-to-end MLOps platform on Crusoe Cloud, leveraging a powerful suite of technologies:

• An early adopter of Crusoe Managed Kubernetes (CMK): Odyssey became a key early adopter of Crusoe’s Managed Kubernetes (CMK) service, helping "kick the tires" on the beta preview. This deep collaboration evolved into a design partnership, where Odyssey's feature requests were fast-tracked and their feedback was instrumental in maturing CMK into a generally available product.
• Dynamic data processing with L40S and autoscaling: For the critical first step of their workflow (processing vast amounts of video and image data) Odyssey utilizes Crusoe’s L40S GPUs with CMK’s autoscaler. This allows them to dynamically burst their data processing pipelines up and down on demand, efficiently preparing their unique datasets for training without having to pre-provision or manage nodes.
• A future-proof fleet of advanced GPUs: Odyssey conducts its large-scale training runs on clusters of NVIDIA H100 and H200 GPUs across Crusoe’s global data centers. Looking to the future, they are among the first to deploy the new NVIDIA GB200 rack-scale solution on Crusoe, ensuring their infrastructure remains at the cutting edge of AI.

The results: Accelerated innovation and massive cost savings

Crusoe enabled Odyssey to move faster, innovate more freely, and achieve significant efficiencies. The results are evident across their business.

• Significant cost savings: By leveraging Crusoe’s ability to provide burst capacity for training and data processing, Odyssey completely avoided the need to provision for peak load. This flexibility has been transformative; had they been forced to commit to peak capacity, their infrastructure spending would have been up to 4.4 times higher.
• Accelerated time-to-market: The close collaboration and Crusoe’s hyper-responsive support directly accelerated Odyssey’s development timeline. Crusoe’s willingness to co-develop features for CMK enabled Odyssey to get their product launched successfully and on schedule.
• Proven scalability and success: With Crusoe’s infrastructure, Odyssey successfully launched its "Explorer" application, smoothly scaling to serve hundreds of thousands of users from its initial research preview.
• Reliable compute for the future: Odyssey is not just a customer; they are a key partner whose feedback directly influences Crusoe’s product roadmap, helping to improve the platform for the entire AI ecosystem. With a clear path to the latest technology like the NVIDIA GB200 NVL72, Odyssey is perfectly positioned to continue pioneering general-purpose world models.

Ready to build the future of AI on a flexible, cost-effective, and collaborative cloud? Learn more about Crusoe Cloud and experience Odyssey's groundbreaking models for yourself.