I am not a programmer, and I don't do AI for a living. I've spent over a year learning about LLMs because I've always liked learning new tech. I make YT videos and X posts for the community of users and developers alike.

What if something similar to OpenAI Stargate used open weights models with distributed compute funded by decentralized tokens on blockchain?.

Confused with all the open source AI lingo?. Give me 12 minutes and I'll break it down. 👇.

OpenAI Stargate will likely show us impressive outcomes and major impact on solving problems. But it's. ▪︎ Funded by a centralized bank.▪︎ Sanctioned by a centralized government.▪︎ Uses closed centralized tech. Now imagine the same thing with open source on blockchain.

This is the research they speak of that will improve the world and the lives of everyday people. • Enhance AI Development.• Secure U.S. AI Leadership.• Create Jobs.• Support National Security.• Expand AI Infrastructure.• Advance Sustainable Energy.• Boost Economic Growth.

How anyone can get started with open source AI. If you're a total beginner curious about it but overwhelmed by the jargon, this video breaks it down simply. No prior knowledge needed!. Topics covered:. • What is an open source LLM.• Base models.• Instruct models.•

I had no idea!. I knew you could make scheduled tasks in Grok 4, but didn't know it has its own JSON!

o3-alpha:. The best facts I can find, avoiding speculation (sources included). • Originally discovered on LM Arena and WebArena testing sites. • Experimental model, identified as "o3-alpha-responses-2025-07-17" and "Anonymous-Chatbot" in testing. Source: BleepingComputer, July

Hermes 3 from @NousResearch is the #1 dataset on @huggingface . I plan to use it. Don't sleep on Nous. They:.- Build amazing open source fine-tunes.- Built distributed compute.- Have well curated datasets.- Are fully transparent with model weights and code.- Post based takes on X

OK, so OpenAI has an internal model that beat a hard math test. Intriguing, but there are some things to unpack:. - General LLM (not specialized).- Reasoning model, probably RL.- Can't do your taxes.- Won't be released w/ GPT-5.- No research paper.- Typical lack of transparency.

When I see posts of people saying "X model is better than Y", I ask "at what?". Just like people have different strengths and weaknesses, so do LLMs. From vibe testing, I've figured out what works best for my needs and so should you. The current frontier: o3, Grok-4 & Flash 2.5

GPT-5 must be close. - An OpenAI employee said it today. - There's a reasoning model on their benchmark repo.- gpt-5-reasoning-alpha-2025-07-13 is in the biology benchmarks

Excited to try my first fine-tune!. I think I have my settings nailed down. Grok helped. Going to use SmolLM3 from @huggingface with Hermes 3 dataset from @NousResearch and @UnslothAI using @ollama. If you're new to it like me, heres's the convo of Grok teaching me:.

Hey @xAI Grok scheduled tasks suck. I love Grok but want this to be better. I asked for a daily update of AI news and model releases in the last 24 hours. It shows stuff from days or months ago. This feature could be fire if it's refined a bit.

Reminder! . While ChatGPT has a new agent, other choices exist. - Manus.- MiniMax.- Kimi K2.- DeepAgent . Some have more transparent output, too. Less gatekeeping means more freedom. Am I forgetting any?.

I bet the ChatGPT agent jailbreaks will be crazy 🤪.

Ofc you'd have to give website credentials to the ChatGPT agent.

Good idea that @OpenAI is rolling out the new agent features before GPT5. Get everyone familiar first, then roll out the rest later. They said a long time ago they would not release everything at once.