Sometimes, poor responses from the model aren’t due to its quality, but because of specific rules within your projects or an overloaded context window. In such cases, starting a new chat can lead to better results. If you add a lot of context — like files, folders, or images — and use a vague prompt, the model might also provide weaker responses because it “tries to grasp everything at once” and loses sight of the core task. Additionally, there’s no real reason for us to switch models — API prices are the same.
