A ChatGPT Pro subscriber reports that Scheduled Jobs in regular ChatGPT no longer behave as before, with model-selection options now matching those in ChatGPT work. The user also reports persistent message-delivery timeout errors and questions whether the feature's background execution has been degraded.
A two-year ChatGPT Plus subscriber reports growing instability across OpenAI's ecosystem after recent Codex, ChatGPT apps, Work, and model changes. They cite the Work agent consuming the shared 5-hour usage bucket, UI bugs in the macOS classic app, inconsistent response quality on Android, and disappearing features like library. The post reflects frustration among Plus users over product fragmentation, with the user reluctant to switch to Gemini or Claude despite considering it.
A two-year ChatGPT Plus subscriber reports growing instability across OpenAI's ecosystem after recent Codex, ChatGPT apps, Work, and model changes. They cite the Work agent consuming the shared 5-hour usage bucket, UI bugs in the macOS classic app, inconsistent response quality on Android, and disappearing features like library. The post reflects frustration among Plus users over product fragmentation, with the user reluctant to switch to Gemini or Claude despite considering it.
A Reddit user using GPT-6 Astra to generate 70+ page documents from 100,000+ characters of source material reports ChatGPT grinding for 30-40 minutes on massive tasks, then failing with a 5-hour usage limit error and no output. The user argues OpenAI should warn before tasks that will breach limits, estimating each wasted run at $3-$5 in API-equivalent compute, and has had to restart shrunken tasks multiple times.
A Reddit user using GPT-6 Astra to generate 70+ page documents from 100,000+ characters of source material reports ChatGPT grinding for 30-40 minutes on massive tasks, then failing with a 5-hour usage limit error and no output. The user argues OpenAI should warn before tasks that will breach limits, estimating each wasted run at $3-$5 in API-equivalent compute, and has had to restart shrunken tasks multiple times.
OpenAI Plus subscribers report that their Codex weekly usage limits are resetting unexpectedly, sometimes multiple times per week, rather than on the scheduled Thursday. The issue appears to be a bug or policy change affecting billing and usage tracking.
OpenAI Plus subscribers report that their Codex weekly usage limits are resetting unexpectedly, sometimes multiple times per week, rather than on the scheduled Thursday. The issue appears to be a bug or policy change affecting billing and usage tracking.
A Reddit user reported that their scheduled tasks in Claude, which were set to run on older models like Sonnet or Opus, have automatically switched to the newer Fable model without their consent. The user is concerned about burning credits on tasks that don't require the advanced model and is seeking a way to control model selection for scheduled tasks.
A Reddit user reported that their scheduled tasks in Claude, which were set to run on older models like Sonnet or Opus, have automatically switched to the newer Fable model without their consent. The user is concerned about burning credits on tasks that don't require the advanced model and is seeking a way to control model selection for scheduled tasks.