Generative AI Video Platform Everlyn Taps Walrus as Defacto Data Layer and Leverages Sui for Payments

The creator-focused video generation platform will leverage Walrus to support hot retrieval of content, audit-read video AI training, and immediately support storage for over 5,000 user-generated clips.

Everlyn, the generative AI platform that turns static images into high-quality video in seconds, has partnered with Walrus to act as the default data layer for Everlyn-1, its fastest pure video generation model that can convert images to video in as little as 16 seconds. Everlyn’s platform will now benefit from Walrus’ cost-effective data management model, onchain auditable records, and high performance batching for small writes to maintain speed while reducing storage overhead.

Walrus will immediately provide data storage for over 5,000 user-generated videos already created on Everlyn, ranging from 480p to 720p HD quality. The partnership will also bring training data sets currently stored on AWS and Azure onto Walrus, enabling real-time model improvements. Additionally, Everlyn will leverage Sui for smart payments and smart contracts, which will be directly integrated into the generative video suite.

“Everlyn is at the forefront of supercharging the creator economy, enabling users around the globe to transform still images into high-quality video clips in seconds instead of hours,” said Rebecca Simmonds, Managing Executive at Walrus Foundation. “This partnership showcases the power Walrus brings for AI use cases, allowing for real-time model improvements and agent memory persistence, with smart contracts and payments provided through Sui’s fast and reliable blockchain network.”

The initial stages of the partnership will see the migration of over 50GB of training datasets, model checkpoints, KV-caches, sparse attention maps, and Agent memory, all stored on the Walrus platform, in addition to all new videos generated by users. Through APIs and SDKs, the partnership will also allow custom integration into ecosystems and communities that currently use Everlyn, allowing it to effectively scale alongside creator communities.

“While centralized cloud services were suitable for our platform during the beta phase, we knew early on that these solutions would price out many creators and be too expensive to maintain long term for high-speed video creation needs,” said Dr. Sernam Lim, Cofounder and CEO of Everlyn. “Walrus offers us a highly performant yet cost-effective solution that allows us to maintain the speed Everlyn has become known for, making this partnership a no-brainer.”

The Everlyn video generation suite offers users significant enhancements over traditional AI video generation models, allowing high-quality videos to be created in seconds as opposed to hours with other models. Available worldwide, Everlyn processes video generation requests significantly faster than platforms like Midjourney or Veo 3.

This announcement follows other recent Walrus integrations, including partnerships with Yotta Labs, TensorBlock and Inflectiv, advancing its broader push into decentralized AI and intelligent data infrastructure. By supporting partners like Everlyn, Walrus continues to expand its role as the foundation for verifiable, programmable data powering the next generation of Web3 applications.

To learn more about Everlyn and how it is powering the fastest pure video generation onchain, visit: https://everlyn.ai