OpenAI dropped its most powerful model yet on Thursday, releasing GPT-5.5, internally nicknamed “Spud,” just one week after rival Anthropic unveiled its own latest model. The back-to-back releases underscore just how quickly the AI race is moving.
What Is GPT-5.5?
OpenAI co-founder Greg Brockman described the new model to reporters as something that represents a genuine step forward, not just an incremental upgrade.
“This is a new class of intelligence. It’s a big step towards more agentic and intuitive computing,” Brockman said.
Compared to its predecessor GPT-5.4, the new model is a “faster, sharper thinker for fewer tokens,” Brockman added. Despite the jump in capability, OpenAI says real-world response speeds remain on par with GPT-5.4. The model can also handle multi-step tasks with far less hand-holding from the user, planning its own approach, using tools, checking its own work and pushing toward a final result.
Where It Shines
OpenAI says the biggest improvements show up in coding, computer use, general office tasks and early-stage scientific research, all areas that demand reasoning across longer stretches of information and sustained execution over time. Rather than walking the model through a task step by step, users can hand it a messy, multi-part problem and let it figure things out.
Teams that got early access reported being able to review thousands of additional documents, catch errors in quickly written code and save up to 10 hours of work per week, according to the company.
Who Can Use It
GPT-5.5 is available starting Thursday through ChatGPT and Codex for paid subscribers. API access is coming soon, once OpenAI finishes adding additional cybersecurity guardrails to the system.
Nvidia’s Role
The model was trained on Nvidia GPUs, continuing a pattern from previous OpenAI releases.
According to a report from Axios, Justin Boitano, Nvidia’s vice president of enterprise computing, said the model can function as a “chief of staff,” supporting AI agents that are already performing employee-level work inside the company. Nvidia also worked with OpenAI to put together a blueprint aimed at making it easier for businesses of all sizes to deploy the technology.
Nvidia says its newest chips cut the cost of running advanced AI like GPT-5.5 by as much as 35 times per token, a figure that matters a great deal for companies trying to scale up AI usage without sending their IT budgets off the rails.
The Bigger Picture
The release comes at a pointed moment for OpenAI. The company’s leadership had reportedly described Anthropic’s growing momentum as a “code red” and a “wake-up call,” prompting a strategic shift toward winning over business customers. GPT-5.5 and its Nvidia partnership appear to be a direct response to that priority.
Brockman also used the occasion to paint a broader picture of where things are heading. “We are moving to a compute-powered economy,” he said, arguing that AI capacity, and the computing power behind it, will increasingly become the engine driving how work gets done.
