@Govindtwtt Better yet, I have deleted its memory completely by moving the .Claude files somewhere off the machine. I miraculously starts working much much better... but you have to give it the context again.
@txmps__@beforegodapp@ClaudeDevs I have been using Grok 4.6 and it's better than the Fable we have had for the last week. Fable is now Opus but less annoying.
@ardnox@sattyyouneed Use a YouTube transcript API to pull transcripts, break them down and decypher all the ins and outs of a strategy that way. But yes it's horrible
@ClaudeDevs one watcher agent that consumed ~607K tokens in status-polling loops producing no output. The coordinator session alone measured ~850 ast turns, ~269M cache-read~12.6M cache-creation ~1.15M output it ate a 200 max account in 3 hours
@ClaudeDevs your update Your update caused all of my subagents that were previously Sonnet to be dispatched via the Agent tool without an explicit model parameter inherited the session model (Fable) instead of Sonnet — including survey agents, workers, and
@ClaudeDevs@claudeai How is this possible with Fable orchestrating and Sonnet Sub agents? My Fable usage never reset when I upgraded. I had fable audit my usage and have submitted a ticket. Help
@ClaudeDevs I upgraded from 5x to 20x after your update and none of my usage limits updated . 1st session after upgrade is at 60% but fable weekly limit is at 75%. Please check this even had fable audit my usage and it said it's off