- Cloudflare has integrated OpenAI’s GPT-5.4 into its Agent Cloud platform, extending frontier model access to millions of enterprise customers.
- The Codex harness is now generally available in Cloudflare Sandboxes, a secure virtual environment for building and testing AI applications.
- Agent Cloud runs on Cloudflare Workers AI, an edge-computing platform designed to reduce inference latency at global scale.
- OpenAI’s APIs process more than 15 billion tokens per minute; Codex had 3 million weekly active users at the time of the announcement.
What Happened
Cloudflare announced on April 14, 2026, that its Agent Cloud platform now includes OpenAI’s GPT-5.4 model and Codex, making them accessible to the company’s enterprise customers in a production-ready deployment environment. The integration allows businesses to build and deploy AI agents that automate tasks such as customer response, system updates, and report generation. The announcement was published on OpenAI’s blog.
Why It Matters
The partnership routes OpenAI models through Cloudflare’s global edge network rather than requiring enterprises to call centralized OpenAI API endpoints separately — a design the companies say reduces latency and simplifies deployment architecture. OpenAI already serves more than 1 million business customers and holds enterprise agreements with Accenture, Walmart, Morgan Stanley, BNY, and others; the Cloudflare integration adds a new deployment surface within an ecosystem many of those organizations already use. Dane Knecht, chief technology officer at Cloudflare, stated in the announcement: “By bringing OpenAI’s powerful models directly into the Cloudflare environment, we are collapsing the distance between intelligence and the end user.”
Technical Details
Agent Cloud is built on Cloudflare Workers AI, the company’s platform for running AI inference at the network edge rather than in centralized cloud regions. The Codex harness — enabling software development automation via OpenAI’s Codex model — is now generally available within Cloudflare Sandboxes, an isolated virtual environment where developers can build, run, and test AI applications. According to the announcement, OpenAI’s APIs process more than 15 billion tokens per minute, and Codex had reached 3 million weekly active users. Cloudflare stated that Codex support will be extended to the Workers AI runtime directly, though no timeline was provided.
Who’s Affected
Enterprise developers building on Cloudflare’s platform can now access GPT-5.4-powered agents without routing inference requests to separate external endpoints, with model execution handled within Cloudflare’s infrastructure. Knecht described the outcome as enabling applications that are “lightning-fast and globally scalable by default, whether it’s using Codex to streamline development workflows or deploying AI applications and agents at scale.” Existing OpenAI enterprise customers — including Intuit, Thermo Fisher Scientific, State Farm, and BBVA — represent an adjacent audience that could adopt Agent Cloud as an additional deployment surface.
What’s Next
Cloudflare said Codex availability will expand from Sandboxes to the broader Workers AI runtime, with no release date specified in the announcement. Rohan Varma, product lead for Codex at OpenAI, framed the integration in terms of production readiness: “Cloud agents are quickly becoming a foundational building block for how work gets done, and with Cloudflare, we’re making it dramatically easier for developers to deploy, production-ready agents powered by GPT-5.4 and Codex to run real enterprise workloads at scale.” Both companies characterized the integration as targeting active production deployment rather than prototyping or experimentation.
