FUTO is introducing “Swipe,” a swipe-typing model designed to run on-device. The available reports describe the project as an open model for swipe typing, emphasizing local inference rather than relying on cloud-based processing. On Product Hunt, the post presents Swipe as an on-device approach and includes a link for discussion, but does not provide detailed performance metrics or deployment information in the material provided here. Hacker News also features the announcement as “FUTO Swipe – A new swipe typing model,” largely focused on community discussion rather than additional technical documentation. Across both sources, the shared theme is that Swipe targets swipe typing using an on-device model and is presented as open. However, the provided excerpts do not specify licensing terms, supported platforms, model size, accuracy, training data sources, or availability timelines. Based on the cited content alone, the key confirmed points are the model’s purpose (swipe typing), its intended execution on-device, and its positioning as open.