OpenAI Finally Found Why Your Codex Usage Was Disappearing!
If your **ChatGPT Codex usage limit** has been disappearing much faster than normal, you weren't the only one noticing it. After days of complaints from Codex users across Reddit and X, OpenAI investigated — and found multiple real efficiency problems affecting Codex usage. In this video, I break down exactly what OpenAI discovered, why some **Codex usage limits** may have been draining faster, how prompt caching and context compaction affect your allowance, what went wrong with image-heavy sessions and Computer History, and why even Codex's conversation-title generator was consuming more usage than intended. And there's an even bigger development: OpenAI says it discovered a separate “novel approach” that could make Codex **significantly more efficient**. Since the original announcement, Tibo Sottiaux has confirmed that the latest **Codex reset has now propagated to accounts** and that OpenAI has already landed some fixes related to the usage problems. He says users should notice a positive difference, with more improvements still coming. So was OpenAI secretly reducing Codex limits? Were Reddit users right? Or was this mostly an efficiency problem happening behind the scenes? That's what we're digging into. This video covers: • ChatGPT Codex usage limits • OpenAI Codex rate limits • Why Codex usage was draining faster • Codex weekly usage and resets • Prompt caching / cache-hit rates • Codex context compaction • Images in long Codex sessions • Computer History usage • Conversation-title generation • OpenAI's Codex efficiency fixes • The latest Codex usage reset • What OpenAI plans to optimize next ━━━━━━━━━━━━━━━━━━ 📌 OFFICIAL / PRIMARY SOURCES Tibo Sottiaux — main Codex rate-limit investigation: [Main Codex usage investigation on X](https://x.com/thsottiaux/status/2091407991736332689?utm_source=chatgpt.com) Tibo — earlier update confirming worse cache-hit rates for some users: [Cache-hit-rate investigation on X](https://x.com/thsottiaux/status/2091033630147854385?utm_source=chatgpt.com) Tibo — reset timing announcement: [Codex reset timing post on X](https://x.com/thsottiaux/status/2091412393368945027?utm_source=chatgpt.com) Tibo — clarification of the reset time: [2 PM clarification on X](https://x.com/thsottiaux/status/2091413240337326588?utm_source=chatgpt.com) Tibo — reset propagated + usage fixes landed: [Latest reset/fixes update on X](https://x.com/thsottiaux/status/2091688655828246890?utm_source=chatgpt.com) Tibo's X account for continuing updates: [Tibo Sottiaux on X](https://x.com/thsottiaux?utm_source=chatgpt.com) ━━━━━━━━━━━━━━━━━━ 📚 OPENAI DOCUMENTATION OpenAI — Using Codex with your ChatGPT plan: [Official Codex usage documentation](https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan?utm_source=chatgpt.com) OpenAI — Unrolling the Codex Agent Loop: This is particularly useful if you want to understand prompt caching, context management and automatic compaction inside Codex. [Unrolling the Codex Agent Loop](https://openai.com/index/unrolling-the-codex-agent-loop/?utm_source=chatgpt.com) OpenAI — ChatGPT release notes / Computer History: [Computer History documentation](https://help.openai.com/en/articles/11391654-chatgpt-team-release-notes?utm_source=chatgpt.com) OpenAI Codex: [Official OpenAI Codex page](https://openai.com/codex/?utm_source=chatgpt.com) ━━━━━━━━━━━━━━━━━━ 🔥 REDDIT THREADS / COMMUNITY REPORTS “My 20x is draining like it's 5x — wtf happened this week?” [Read the Reddit discussion](https://www.reddit.com/r/codex/comments/1vtu0iq/my_20x_is_draining_like_its_5x_wtf_happened_this/?utm_source=chatgpt.com) “Wtf is going on with the usage limits??” [Read the Reddit discussion](https://www.reddit.com/r/codex/comments/1vtth3d/wtf_is_going_on_with_the_usage_limits/?utm_source=chatgpt.com) “There's something wrong with limits” [Read the Reddit discussion](https://www.reddit.com/r/codex/comments/1vpl0g5/theres_something_wrong_with_limits/?utm_source=chatgpt.com) One user's detailed analysis asking whether Pro 5x weekly Codex usage had effectively dropped ~44%: IMPORTANT: This is community analysis, NOT an OpenAI-confirmed 44% reduction. [Read the Reddit analysis](https://www.reddit.com/r/codex/comments/1vu3u05/did_pro_5x_weekly_codex_usage_drop_44_after_the/?utm_source=chatgpt.com) “What are the actual hard numbers on the limits problem?” A useful counterpoint discussing how difficult it is to separate genuine limit changes from model choice, caching, context and workflow differences. [Read the hard-numbers discussion](https://www.reddit.com/r/codex/comments/1vugoc4/what_are_the_actual_hard_numbers_on_the_limits/?utm_source=chatgpt.com) Codex Usage & Operation discussion / megathread: [Codex usage megathread](https://www.reddit.com/r/codex/comments/1vqwcbs/codex_usage_and_operation_discussion_updated/?utm_source=chatgpt.com) #openai #codex #ai
