Tell HN: OpenAI brings back 5 hour limit for plus and business standard users — SkimNews
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- OpenAI reinstated a 5-hour session rate limit after a period of only weekly limits applied, prompting users to report hitting session caps and consider upgrading, buying another plan, or moving off Codex.
- The 5-hour window begins when a user sends their first token after extended inactivity, and on the $20/month plan consumes roughly 15% of the weekly allowance per a commenter.
- Per-model usage multipliers introduced in early August replaced OpenAI's prior flat $60-of-usage-for-$10 deal, with different models now yielding $15, $30, $60, or $100 in monthly usage value.
- Commenters characterize the constant policy shifts as "bait & switch," saying tracking which model offers the best capability-to-cost ratio requires ongoing mental effort they don't want to expend.
- One $200/month subscriber describes using OpenAI simultaneously as a project manager, executive assistant, business analyst, software developer, and international accountant, calling the value strong.
- A commenter proposes OpenAI offer an unlimited $500–$1,000/month tier capped at 10 concurrent connections to enable 24/7 agentic workflows without rate-limit friction.
Why it matters: The 5-hour cap squeezes lower-tier subscribers hardest — $20/month users lose ~15% of weekly allowance per window — while per-model usage multipliers force power users to constantly recalculate which model offers the best dollar value, adding friction as rival services like Hyper compete at $20/month.
Ask SkimNews
