Cloudflare Integrates OpenAI GPT-5.4 and Codex into Agent Cloud

Cloudflare has added OpenAI’s GPT-5.4 and Codex models to its Agent Cloud, giving enterprises a hosted environment to build, deploy, and scale agentic workflows. The newly reported integration positions Agent Cloud as a place where developers and teams can run agents backed by OpenAI’s latest models while relying on Cloudflare’s infrastructure and orchestration.

Operationally, the integration means teams can assemble agents that use GPT-5.4 or Codex inside Cloudflare’s agent runtime and deploy them at scale without separately provisioning model hosting. The companies present the pairing as focused on speed and security, suggesting the offering is aimed at production use rather than exploratory experimentation. The likely immediate users are application developers, product and automation teams, and infrastructure decision-makers evaluating where to run agentic tooling.

What changes in practice is a lower-friction route to operate model - driven agents: fewer integration touchpoints and a single provider surface for runtime and orchestration. That can accelerate deployment timelines, but it also centralizes operational questions-observability, access management, data residency, and auditing-within the hosting environment. The announcement does not include details on pricing, SLAs, or specific security controls, so organizations will need to vet those elements against compliance and reliability requirements before migrating critical workflows.

Watch for release notes, documentation, early customer examples, and any follow-up on governance or control features. Those signals will determine whether the integration simply simplifies sandbox experiments or meaningfully shifts how enterprises choose to run agentic automation at scale.