I believe these labs really nailed the restrictions on independence for these models. I have seen collegues coding bs with fable 5. Genuinely, the skill issues is feel by the agent and the output is alike