OpenAI Codex Head Tibo Confused by Pro User Voting Results
OpenAI Codex head Tibo expressed confusion over the results of a vote among Pro users, which he expected would show a preference for 2x usage in the Pro 100/200/500 plans, even at the cost of reinstating a 5-hour limit to smooth the demand curve.
The vote asked Pro users about their preferences, weighing usage multiples against limit structures. Tibo noted that the results did not clearly favor the 2x option as he had anticipated.
Previously, OpenAI had promised not to reintroduce the 5-hour limit in certain Pro plans to allow users more flexibility in using their weekly quotas, while also adjusting the usage calculation method, changing the value of API costs corresponding to the same subscription price.
This vote occurred against the backdrop of ongoing adjustments to the usage limits for Codex and ChatGPT Work, with community discussions focusing on the impact of the 5-hour limit on high-intensity usage patterns. Some users indicated that reinstating the limit would hinder productivity and consider switching to other platforms.
Tibo's comments reflect the product team's attempt to smooth the computing demand curve through limit design, while user feedback emphasizes the importance of flexible usage without the 5-hour restriction, rather than simply increasing multiples.
In subscription-based AI products, such votes serve as tests of demand-side preferences, influencing subsequent adjustments to limits and usage policies. High-usage users prefer plans without short-term limits, and the platform must balance computing smoothness with user experience.
OpenAI has previously managed peak demand by resetting quotas or adjusting default limits, and the results of this vote may impact the decision on whether to reintroduce the 5-hour limit.
Source: Public Information
ABAB AI Insight
OpenAI, through Codex head Tibo, continues to publicly test Pro users' preferences regarding usage limits. Historically, it has removed or postponed the 5-hour limit to attract high-intensity users while managing computing load through subscription resets and usage calculation adjustments, making multiple changes to the Pro 100/200/500 plan quota structure during 2026.
The resource path allocates computing costs through limit and multiple designs across different subscription tiers, aiming to smooth the demand curve, reduce peak pressure, and maintain the attractiveness of high-priced plans. The voting results indicate that users are more resistant to short-term limits disrupting their usage rhythm rather than simply pursuing higher multiples.
This is similar to competitors like Anthropic in their Claude series, which compete for heavy users by resetting limits and adjusting windows, and is akin to historical cases of SaaS products repeatedly adjusting between unlimited usage and hard limits; OpenAI is currently in a testing phase transitioning subscription usage from flexible weekly quotas to potentially reintroducing short-term constraints.
Essentially, this involves the intersection of pricing power transfer and technical constraints: the platform attempts to control peak computing demand through limit structures, driven by inference costs and users' high-intensity usage behaviors, leading to a divergence between users' preference for continuous usage rhythms and the platform's goal of smoothing the curve.
ABAB News · Cognitive Laws
- Multiples can double, but rhythms cannot be interrupted.
- Limit = Computing Smoothness × User Attrition.
- Voting asks for preferences, results measure tolerance.