The Oracle of AI
There are a TON of SOTA LLMs out there (State of the Art Large Language Models - long word, I know!).
It's easy to get lost trying to figure out which one is right for you, so let me help you out.
But first, let's quickly explain what "SOTA LLM" actually means:
A SOTA LLM is considered to be among the most advanced and highest-performing Large Language Models you can find right now. It's basically the cutting edge of language AI.
Let's break that down:
State-of-the-Art (SOTA): In research and tech, this term means the best current method, model, or result for a specific task or problem. Think of it as the benchmark everyone is trying to beat. Just remember, SOTA is always changing as things get better and better!
Large Language Model (LLM): These are AI models trained on HUGE amounts of text and code. They're designed to understand, create, and work with human language. "Large" refers to the massive scale of their parameters (those numbers inside the model that are learned when it's trained) – we're talking billions!
So, in short, a SOTA LLM is the absolute top of the line in LLM tech right now. It's a model that shows amazing performance, super advanced abilities, and often uses really innovative stuff. Just keep in mind, this field moves fast, and what's SOTA today might be old news tomorrow!
Okay, now that we're all on the same page about SOTA LLMs, let's dive into my LLM stack.
After reading this, you'll know more about LLMs and how to use them to seriously boost your work. (This is Part 1 – Part 2 is coming next week!)
My Top 3 SOTA LLMs (Part 1):
Claude Sonnet (Anthropic)
This is, in my opinion, the most sophisticated model out there right now. I have no idea how Anthropic does it, but they're definitely onto something amazing, and it makes me happy to use! Sonnet 3.5 just got upgraded to 3.7!
Why I love it: Because it's so smart, most people use Sonnet as a code assistant. While pretty much all LLMs can spit out code (some LLMs need to be told "give me the code in a [code language] code block"), Sonnet really shines at writing and basic research/brainstorming. It's also fantastic at generating graphs and charts. Think great overviews, reports, and executive summaries.
Bonus: You can feed it documents, GitHub links, and images! Ask it questions about them, or give Claude info that it might not have in its training data. Super useful!
ChatGPT-4 Turbo (OpenAI)
My absolute favorite all-rounder. It can basically do everything, and like the name says – it's FAST! I use GPT-4 Turbo almost every day to quickly grab info on books, people, places, and historical events. I can ask it basic questions, upload docs and images – basically everything Claude can do.
Coding: GPT-4 isn't the best for coding overall, but it's seriously good at Python!
Why I love it (vs. Claude): GPT-4's responses are super on-point and not too fluffy. Claude yaps a lot, which is great when you want to really dig into complex topics, but sometimes you just need a straight answer, right?
Think of Claude as the master of Education, while GPT-4 Turbo is more the Detail-Oriented Analyst
Sonar Pro (Perplexity.ai)
Have you ever used an AI-powered web-browsing LLM? You NEED to! It's a massive time-saver and completely changes how you interact with information.
Web Browsing Power: You don't just get URLs – your response to your question is way more in-depth. You can ask follow-up questions and even get explanations of the information you're looking at while you're looking at it.
Summaries & Reports: It'll give you a summary of your request and even write up a whole report for you if you ask!
Check it out: They have a pretty generous free tier, so you can try it out for yourself » here
Stay Tuned for Part 2! I'll be writing more about features, functions, tips, and tricks in future posts, so stay tuned! If you want to save time & energy, grow your skillset, and master AI – you're in the right place.
Want to Get Really Ahead of the Game? I've got a physical magazine for you! It arrives at your doorstep, updated every month. Check it out!
myllm.news
Be back next week!