WIKI // AI CODING TROUBLESHOOTING
AI Coding Rate-Limit & Cost Error Wiki
Hit an unexpected usage cap or sudden cost spike? Pick your error message below for the root cause, recovery calculation, and step-by-step fix.
- Claude Code / Claude Pro / Max
You’ve reached your 5-hour usage limitClaude Code "5-Hour Usage Limit Reached": Recovery Time & Fixes
Claude enforces a 5-hour rolling session window starting from the first message of the active cycle—not 5 hours after the rate-limit error appeared.
View fix & calculator → - Claude Pro / Max (7-Day Rolling)
Weekly usage limit reachedClaude Pro / Max "Weekly Usage Limit Reached": Pacing & Alternatives
Claude’s 7-day weekly limit is enforced independently of the 5-hour session cap; once exhausted, you must wait for your personal weekly reset or fall back to pay-as-you-go API.
View fix & calculator → - OpenAI Codex (ChatGPT Plus / Pro)
Codex usage limit exceededOpenAI Codex Usage Limit Exceeded & Reset Card Troubleshooting
When Codex hits a usage cap, check whether the latest announcement is an all-user global reset or a bonus Reset Card, as cards often target specific plans or require manual redemption.
View fix & calculator → - Claude / OpenAI API & Coding Agents
Low Prompt Cache Hit Rate / Cost SpikePrompt Cache Misses & Sudden AI Coding Cost Spikes: How to Fix
Healthy agentic coding sessions hit 80%–95% prompt cache reads. When dynamic system prompts, mid-session model switching, or expired cache TTLs break prefix matching, costs surge 5x–8x.
View fix & calculator → - Cursor Pro (500 Fast Requests / 30d)
Fast requests limit reachedCursor 500 Fast Requests Exhausted: Monthly Pacing & API Math
Cursor Pro’s 500 fast requests average ~16 per day over a 30-day cycle. Heavy multi-file Agent mode early in the month quickly burns through them; use our 30-day planner to pace daily calls.
View fix & calculator →
