OpenAI’s CEO Sam Altman described GPT-4.5 as a “giant, expensive model” and noted GPU shortages as a key challenge in the phased rollout. The company plans to add tens of thousands of GPUs to meet demand.