Nvidia and Meta Platforms agreed in October 2026 to expand their work on Meta's Muse AI agent and Muse family of models to improve inference and training performance