CoreWeave Forge targets GPU use in AI post-training

CoreWeave has launched CoreWeave Forge, a full-stack AI cloud connecting model deployment, evaluation, and improvement. Its RL Rollouts capability supports repeated cycles of generating training responses and updating model weights via peer-to-peer transfer rather than cold loads from object storage. CoreWeave AI Object Storage now supports cross-region writes to keep GPU utilization high between training rounds. In partnership with You.com and Nvidia, the companies post-trained Nemotron 3.5 Lightning in eight hours using You.com's web search tools, demonstrating that continuous post-training is no longer limited to frontier AI labs.