Builds open-weight frontier models and the software and infrastructure needed to customize and deploy them. Its strategy combines large-scale pretraining and reinforcement learning with publicly released weights, research and development tools.
See something inaccurate or outdated?
