I’ve been getting much better GPT-4.5 responses by making it reason. You can use XML tags like <thinking> and <answer> to break up its reasoning from the final answer and improve its responses. Here’s a starting template: https://t.co/JifIoChBsg

i have noticed that LLMs like claude and gpt-4o work really well with this prompt, it instructs them to 'contemplate' for a bit before giving the final answer. https://t.co/9o5Q9M2Atv