Currently Pickaxe caps Grok Output at 30k output tokens per response. This limitation is not on xAI’s end. Grok doesn’t limit the output tokens.
There are some Claude models I only use because I need 128k output tokens. I would switch those models to Grok 4.1 in a heartbeat if Grok had an output limit of 128k or higher.
The same output token limit is on Grok 4.20 by the way. I’m very curious to see if Grok 4.20 can draft book length works on its own.