🔙 Throwback Thursday for 2025-02-06

Throwback to January 2025: the costly drama of distillation vs. hyperscale AI models!

🙏 A big thank you to our new sponsor, NexusTek! 🙏

Thank you NexusTek for sponsoring!

Welcome to the latest Hot Fudge Daily. Each day of the week you get thematic value.

• 🌶️ Spicy Edition Sunday FREE (Hot Fudge Sunday weekly newsletter captures the best of Hot Fudge Daily and the most clicked links for the week.)
• 📈 Markets Monday $ (Hot Fudge Daily: What is moving markets and why does it matter for the week ahead?)
• 🔥 Hot Takes Tuesday $ (Hot Fudge Daily: Get fast unfiltered thoughts on news of the day with just a dash of snark.)
• 🤪 Wacky Wednesday $ (Hot Fudge Daily: Catch up on the wild and wooly memes from a variety of deep web #random channels.)
• 🔙 Throwback Thursday $ (Hot Fudge Daily: Examine recent news in a thought provoking Shot and Chaser format.)
• ✅ Final Thoughts Friday $ (Hot Fudge Daily: A clear eyed look back on the week that was and could have been.)
• 🔮 Sneak Peak Saturday $ (Hot Fudge Daily: Steady and sober unpacking of the slow news day where muted press releases and disclosure stories try to get buried before the weekend takes all the attention.)

🙏 And a big thanks to all our sponsors! 🙏(Scroll to the end…)

One headline for a Thursday Throwback caught my eye.

Techmeme: Stanford and University of Washington AI researchers claim they trained AI reasoning model s1, distilled from a Gemini 2.0 model, for under $50 in cloud compute (Maxwell Zeff/TechCrunch)

By Maxwell Zeff / TechCrunch. View the full context on Techmeme.

Back in the ancient AI times of January 2025, the distillation meme was causing lots of hand wringing but probably cost more than $50 to unseat hyperscaler-size trained models.

Techmeme: The Allen Institute for AI releases Tulu 3 405B, an open source model that it claims outperforms DeepSeek V3 and OpenAI's GPT-4o on certain benchmarks (Kyle Wiggers/TechCrunch)

By Kyle Wiggers / TechCrunch. View the full context on Techmeme.