NVIDIA and OpenAI have launched GPT-5.5, a new frontier AI model. The model powers OpenAI’s Codex application on NVIDIA GB200 NVL72 rack-scale systems. This hardware infrastructure delivers a 35x reduction in cost per million tokens. The system also provides a 50x increase in token output per second.
NVIDIA has deployed the GPT-5.5 application to more than 10,000 employees. The company reports significant productivity gains in software development from this rollout.
OpenAI has committed to deploying more than 10 gigawatts of NVIDIA systems for its next-generation infrastructure. This commitment involves millions of NVIDIA GPUs. The move secures a long-term partnership between the two companies.