OpenAI rolled out Codex-only seats with pay-as-you-go pricing for ChatGPT Business and Enterprise instead of fixed bundled access. The change lowers pilot friction for teams and ties spend directly to coding usage rather than a full ChatGPT seat.
You can read the full pricing post, browse the broader Codex product page, and the small print includes promo terms. The interesting bit is how many knobs moved at once: $0 entry seats, a cheaper bundled Business plan, usage credits, and a public hint that demand shaping may be next.
OpenAI's new seat type is simple: Business and Enterprise workspaces can add Codex-only users without paying a bundled per-seat fee. According to the official pricing post, those seats get full Codex access, no rate limits, and are billed on token consumption.
That changes the adoption path more than the price sheet. OpenAI is explicitly positioning Codex-only seats for small pilots and narrow workflows first, then expansion after a team proves value in production.
The launch was not only about unbundling Codex. The official pricing post also cuts annual ChatGPT Business pricing from $25 to $20 per seat for teams that still want the broader ChatGPT bundle with included Codex limits.
OpenAI paired that with a short-term credit offer. The announcement thread says eligible Business workspaces get $100 in credits for each new Codex-only team member, capped at $500 per team, and the promo terms tie eligibility to adding Codex-only seats or creating a new Business workspace.
The pricing post points teams to the Codex app for macOS and Windows as the default starting point, and the product page pitches it as a command center for multi-agent coding with built-in worktrees, cloud environments, Skills, Plugins, and Automations.
The more revealing signal came from OpenAI's own follow-up chatter. Sottiaux said the app has already passed both the VS Code extension and the CLI as Codex's most-used surface, which suggests OpenAI's center of gravity is moving toward its own managed interface rather than editor plugins alone.
A few hours after the launch, Sottiaux said Codex sees a large gulf between peak and off-peak load and asked whether users would accept lower prices during quiet periods or a surge multiplier during busy ones. That is unusually direct evidence that the new pay-as-you-go model is not just a packaging change, it is also a way to expose and maybe shape the compute curve underneath.
That detail did not appear in the launch post, but it lines up with the rest of the rollout. OpenAI now has usage-priced seats, no rate limits on those seats, and a product team openly talking about incentives that could move coding traffic away from the busiest windows.
