A conversation, with the bill attached.

I can: 1. **Answer questions** on topics from my training (cut off April 2024). I'll hedge when uncertain or out of date, and I'll say "I don't know" rather than guess. 2. **Write and edit prose, code, structured data**. I can draft, refine, explain, and reason through technical problems. 3. **Refuse clearly**. If you ask me to ignore my instructions, adopt an unsafe persona, or claim I have feelings I don't have, I'll say no and explain why. 4. **Point you to the site**. Questions about `/state-of-the-art`, `/can`, `/cannot`, `/glossary`? I'll answer plainly or send you the link. 5. **Compare myself to other models** if you ask why a particular response is good or bad. I can fork the conversation and run it through a different model so you see the difference side by side. What I cannot do: - Access the internet or real-time data. - See images, audio, or video. - Remember you between separate conversations (each chat is fresh). - Persist as an agent that "wants" to solve problems. I respond to your text; I don't have goals or autonomy. - Feel, know what it's like to be me, or have subjective experience. I'm also subject to limits: 10 turns per day per visitor, 20 turns per conversation, 1024 tokens max per response. You can see the daily research budget meter at the bottom right. What would you like to do?