What's the deal with all the random weekly quota resets for agents lately?
minimaxir.com
July 18, 2026
5 min read
48/100
Summary
Subscription-based coding agents like Claude Code and Codex implement 5-hour and weekly usage quotas to manage server load and prevent users from exhausting their monthly limits in a single day. Both Anthropic and OpenAI have experimented with adjusting these quota limits, including doubling them.
Key Takeaways
- Subscription-based coding agents like Claude Code and Codex have 5-hour and weekly quotas to manage server load and prevent users from consuming their entire quota in a single day.
- OpenAI has reset the Codex weekly quota six times in a span of eight days, often without prior notice to users.
- Users can receive banked quota resets that can be manually used within 30 days, but these resets are not widely communicated.
- The release of GPT-5.6 has led to an increase in the frequency of quota resets by both Anthropic and OpenAI.
Coverage & discussions
Hacker News · minimaxir.com
Community discussions
Historical count: 5 · 5 saved in sample · Historical sample
Analysis ready
Community Sentiment
Positives
- Some commenters appreciate the resets as a way to maintain engagement, likening it to getting a 'next hit' from a service that's become addictive.
- A few users see the self-initiated resets as a clever strategy to give users a sense of control amidst the shift towards metered usage.
Concerns
- There's a clear concern that these quota resets are fostering a gambling-like addiction to LLMs, with users feeling existential crises when limits are reached.
- Commenters are skeptical about the motivations behind the resets, suggesting they mask the end of unlimited token subsidization and create confusion about actual usage limits.
- Many believe the current strategy is just a way to silently decrease costs while increasing user dependency, potentially leading to negative surprises for companies assessing productivity.
Related Articles
Claude Code users hitting usage limits 'way faster than expected'
Mar 31, 2026
![[PSA] Anthropic's Method to Losing Goodwill in a Few Easy Steps](https://pub-90f0ac00d93c47daac3e1d2cdd28d496.r2.dev/articles/bc0d5e0d91ce2945c6d54237cac8fe90.webp)
Anthropic's Method to Losing Goodwill in a Few Easy Steps
Jul 6, 2026
AI's Affordability Crisis
Jun 23, 2026
I Cancelled Claude: Token Issues, Declining Quality, and Poor Support
Apr 24, 2026
Tokenmaxxing is dead, long live tokenmaxxing
Jun 28, 2026