I’m curious about this as well, especially since all coding assistants I’ve used truncate long before 1M tokens.