OpenAI’s latest frontier model, GPT‑5.5, is now powering Codex—its agentic coding application—on NVIDIA’s GB200 NVL72 rack‑scale systems, marking a major leap in enterprise‑scale AI performance. According to the document, more than 10,000 NVIDIA employees across engineering, product, legal, marketing, finance, sales, HR, operations and developer programs received early access to the upgraded Codex platform, reporting what one engineer described as “blowing my mind” results.
The integration of GPT‑5.5 with NVIDIA’s latest infrastructure has delivered measurable improvements in developer workflows. The document notes that the GB200 NVL72 system provides 35x lower cost per million tokens and 50x higher token output per second per megawatt compared with previous‑generation systems, enabling frontier‑model inference at enterprise scale.
As a result, debugging cycles that previously took days are now completed within hours, while experimentation that once required weeks is being compressed into overnight progress. Teams are also shipping end‑to‑end features directly from natural‑language prompts with greater reliability and fewer wasted cycles than earlier models.
NVIDIA founder and CEO Jensen Huang encouraged employees to embrace the new capabilities, writing in a company‑wide email: “Let’s jump to lightspeed. Welcome to the age of AI.”
To support secure enterprise deployment, NVIDIA IT rolled out dedicated cloud virtual machines for every employee, enabling Codex agents to operate in isolated sandboxes with full auditability. The document explains that the Codex app uses remote SSH connections to approved VMs, ensuring agents can work with real company data without exposing it externally. A zero‑data‑retention policy governs the deployment, and agents access production systems with read‑only permissions.
The rollout reflects more than a decade of collaboration between NVIDIA and OpenAI. Their partnership began in 2016, when Huang personally delivered the first NVIDIA DGX‑1 supercomputer to OpenAI. Since then, the companies have collaborated across the full AI stack, culminating in the joint bring‑up of the first GB200 NVL72 100,000‑GPU cluster, which completed multiple large‑scale training runs and set new benchmarks for system‑level reliability.
According to the document, GPT‑5.5 is the direct result of this next‑generation infrastructure running at full strength—marking a new phase in the evolution of enterprise AI and agentic development.
