OpenAI launches a $100/month ChatGPT Pro subscription, which offers 5x more Codex usage than Plus; the $200/month Pro plan offers 20x higher limits than Plus
OpenAI is launching a revamped $100/month ChatGPT subscription aimed at Codex users. Here's how it compares to OpenAI's existing plans and what it includes.
Context & Ripple Effects
ChatGPT’s paid model began with a $20 Plus pilot focused on reliability and early features. The product’s monetization is now being tied more directly to developer-agent capacity after Codex was introduced to paid business and premium users.
Codex is a meaningful usage driver: OpenAI reported that weekly Codex use had more than tripled early in the year. That makes differentiated usage ceilings more consequential than a conventional feature bundle for users whose work depends on the agent.
First-order effects
- Codex-heavy ChatGPT customers can choose substantially higher monthly usage allowances at $100 or $200, rather than relying on the Plus-tier limit.
- OpenAI creates a clearer price ladder around its most compute-intensive coding workflow, separating occasional users from customers willing to pay for sustained agent access.
Second-order effects
- Higher tiers give developers and small teams a reason to compare ChatGPT plans by effective coding capacity and continuity of access, not simply by model availability.
- Rival coding-agent providers and AI subscription products face pressure to make their own limits, overage policies, and high-usage pricing easier to evaluate for power users.
Third-order effects
- If agent usage continues to concentrate among intensive users, consumer AI subscriptions are likely to evolve toward metered-capacity segmentation, with premium prices reflecting compute demand more explicitly.
- This is part of a broader test of whether AI vendors can turn high-value task automation into recurring revenue without making usage limits too restrictive for professional adoption.
The trend: AI assistants are shifting from broad, feature-led subscriptions toward tiered pricing based on the compute capacity required to complete valuable work.