OpenAI Launches $100/Month Codex Pro 5X Plan After Token-Based Metering Guts Business Tier
OpenAI has restructured Codex access with a new $100/month Pro 5X plan offering 200-1,000 GPT-5.4 local messages per 5-hour rolling window — ten times the Plus plan’s 20-100 messages through May 31, 2026. The $200 ChatGPT Pro plan remains at 20x Plus limits.
The new tier structure
All Codex limits are measured against a 5-hour rolling window shared between Local Messages (tasks run on the user’s machine) and Cloud Tasks (tasks run on OpenAI’s infrastructure):
| Plan | GPT-5.4 msgs/5h | GPT-5.3-Codex msgs/5h | Cloud tasks/5h | Code reviews/week |
|---|---|---|---|---|
| Plus ($20) | 20-100 | 45-225 | 10-60 | 10-25 |
| Pro 5X ($100) | 200-1,000 | 300-1,500 | 100-600 | 100-250 |
| Pro ($200) | 400-2,000 | 600-3,000 | 200-1,200 | 400-1,000 |
The $100 plan is running at 10x Plus limits (not 5x) through May 31 as a launch promotion. It also carries exclusive access to GPT-5.3-Codex-Spark, currently in research preview, with its own dynamic usage limit.
Token-based metering is the real story
On April 2, OpenAI switched Codex pricing from message count to API token-based rates. The change hit Business accounts hardest. Developers report that identical workloads that previously consumed a normal share of the 5-hour limit now drain it ten minutes into a session — a difference they describe as 2.5-5x worse throughput in practice, not the 60% degradation OpenAI’s official framing implied.
The structural reason: Business tier usage is now metered by reasoning time, not request count. Long-context coding tasks that previously counted as one message now consume proportionally more of the rolling quota based on tokens generated. Enterprise accounts without flexible pricing carry the same per-seat limits as Plus for most features.
One developer analysis posted to OpenAI’s community forum summarised the strategic logic: “They effectively killed Plus and Business as viable Codex plans while offering a more expensive alternative — Pro 5X for $100 — that actually provides a strong user experience for power users. They couldn’t raise the price of Plus and Business directly, because many people subscribe for reasons unrelated to Codex.”
5-hour windows break continuity
The rolling 5-hour quota is the design choice generating the most developer friction. A long coding session that hits the limit mid-task forces either a pause, a switch to a lower-capability model, or an upgrade to the Pro tier. Unlike monthly caps, which allow users to front-load or spread usage, the rolling window enforces a fixed throughput ceiling that restarts every 5 hours regardless of how tokens were distributed.
OpenAI’s developer documentation notes that message volume “varies based on the size and complexity of your coding tasks, and where you execute tasks” — meaning the published ranges are soft targets, not hard ceilings, and heavier tasks consume more quota per interaction.
Context
The restructure tracks a broader trend: Anthropic repriced Claude Managed Agents at $0.08/hr in March; GitHub reversed a similar per-minute runner charge for self-hosted Actions in late 2025 after developer backlash. OpenAI’s metering change differs in that it has not been reversed, and the company has not issued an official response to community threads documenting the Business tier degradation. The $100 Pro 5X plan went live April 9.