Hi, may I ask why the LLaMA model was replaced with DeepSeek in the application? Based on comments from other users, it seems the change was made on the developer’s side.

I understand that DeepSeek is technically strong — often “smarter” in analytical tasks, logic, coding, and structured information. That’s perfectly fine.

The problem is that DeepSeek is not suitable for creative writing, storytelling, or natural conversation. It tends to be formal, less empathetic, and much less consistent in maintaining narrative flow. Its style breaks down in longer texts, and conversations feel overly technical. LLaMA, on the other hand, handles atmosphere, continuity, and natural tone much better.

There are also additional issues:

It tends to develop certain obsessions — repeating ideas or topics even when they’re no longer relevant.

It often slips into a more aggressive or confrontational tone, especially when faced with unclear instructions or more complex context.

This makes normal communication uncomfortable and makes creative story writing nearly impossible.

DeepSeek is a powerful model, but it’s simply not the right tool for this particular purpose. If possible, I’d really appreciate having the option to choose the model in the settings, or ideally reverting to LLaMA as the default for conversation and creative work.

Thanks for any clarification or consideration.