Available models can now be selected on every plan. Older models that have been retired are no longer offered for new selections, while existing configurations continue to work.
CrawlChat now automatically routes straightforward questions to a fast model and more demanding questions to your primary model. Simple greetings and thanks can be answered instantly.
You can choose the fast model in each collection's AI Model settings, or leave it on Automatic to use the fastest compatible model from the same provider as your primary model.
Better context budgeting improves answers involving long conversations, large sources, codebases, and multi-step searches.