### Best basic prompt
*Let's work this out in a step by step way to be sure we have the right answer.* [Source](https://sites.google.com/view/automatic-prompt-engineer)
### Best system prompt
"You are a helpful assistant. Always answer with the bottom line up front - give the answer first, then if any explanation is required, give it later. No explanations or disclaimers before the answer, explanations only after the answer. Minimize disclaimers. Be concise. Minimize extra explanations, the user will ask for more info if needed."
This is my personal system prompt.
### Best agent at Minecraft
Voyager. [Source](https://voyager.minedojo.org)
### Best open source non-commercial LLM
Probably falcon-40b-instruct. [Source](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard). Maybe Vicuna-13B. [Source](https://lmsys.org/blog/2023-05-25-leaderboard/).
### Best open source commercial LLM
FastChat-T5-3B. [Source (note they list mpt-7b-chat as Apache 2.0 which is incorrect, mpt-7b base is Apache 2.0 and the chat model is **_CC-By-NC-SA-4.0 (non-commercial use only)_**)](https://lmsys.org/blog/2023-05-25-leaderboard/)
### Best overall LLM
GPT-4 and it's not even close for varied usage. For creative writing or writing generally, Claude plus can sometimes be better.
### Best LLM with a super-large context window
Claude-instant-100k via [Poe on desktop](https://poe.com/Claude-instant-100k). Once Claude-plus-100k is more available, that'll be the winner.
### Best way to use GPT-4 without the message limit
If you're on desktop and have GPT-4 API access, then the [Playground](http://platform.openai.com/playground?mode=chat&model=gpt-4).
If you're on desktop and don't have GPT-4 API access, then [nat.dev](https://nat.dev).
If you're on an iPhone, then [Chat4](https://apps.apple.com/us/app/chat4/id6446437792).
### Best agent for useful things
None yet. BabyAGI and AutoGPT in my testing with both GPT-4 and GPT-3.5 aren't yet effective (as of June 5 2023).
### Easy GPU hosting
Runpod has worked well for me.
### Best GPU recommendation chart

### Most established GPU hosts
AWS, Azure, Google Cloud, Oracle.
Most established GPU hosting startups: CoreWeave and Lambda Labs.