A large-scale multimodal model that processes both text and images, delivering responses that reflect its substantial parameter count. As a mixture-of-experts architecture with 241B total parameters but only 28B active at inference, it balances capability with computational efficiency. It handles visual understanding tasks alongside text reasoning, though deploying it still requires meaningful infrastructure.