Navigate Select ESC Close

Frontier AI On Your Laptop.

2026-06-18 Science & Technology
1.3k
105
11
Prompt Engineering
Prompt Engineering
245.0k subscribers

Unlock all features

FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.

Description

In this video, we will look at DS-4, that enables DeepSeek V4 Flash to run on a 128GB unified memory. LINK: https://github.com/antirez/ds4 Blogpost: https://engineerprompt.ai/writing/dwarfstar-4/ My voice to text App: whryte.com Website: https://engineerprompt.ai/ RAG Beyond Basics Course: https://prompt-s-site.thinkific.com/courses/rag Signup for Newsletter, localgpt: https://tally.so/r/3y9bb0 Let's Connect: 🦾 Discord: https://discord.com/invite/t4eYQRUcXB ☕ Buy me a Coffee: https://ko-fi.com/promptengineering |🔴 Patreon: https://www.patreon.com/PromptEngineering 💼Consulting: https://calendly.com/engineerprompt/consulting-call 📧 Business Contact: [email protected] Become Member: http://tinyurl.com/y5h28s6h 💻 Pre-configured localGPT VM: https://bit.ly/localGPT (use Code: PromptEngineering for 50% off). Signup for Newsletter, localgpt: https://tally.so/r/3y9bb0 00:00 Can It Run Locally 01:42 Why Big Models Don’t Fit 02:39 Quantization Limits Explained 04:26 Dwarf Star Selective Quant 07:56 Calibration and Validation 09:42 SSD Streaming Ends RAM Cliff 15:34 Benchmarks Demo and Wrap

Top Comments (10)

@nangld 2026-06-19

Reminds me of these DOS Overlay Managers in the era of 640k memory limit.

50 3 replies
@MeinDeutschkurs 2026-06-18

Yeah, and please show us the similar GLM project 🎉

34 5 replies
@piotrnowakowski2271 2026-06-21

Massive stretch on "It can run on your laptop".

33 3 replies
@engineerprompt 2026-06-18

I am starting a blog that will cover some more ideas from the videos, written by me and claude (visual etc.): Here is the first one on DrawfStar (DS4): https://engineerprompt.ai/writing/dwarfstar-4/

19 1 replies
@justcruisin109 2026-06-19

This was very interesting so thanks for taking the time to put it together (as well as the graphics). Given that 128GB of unified memory is a more practical target architecture (although I would love the 512GB option - $$$$$) this project offers the opportunity to run high quality models locally that otherwise would be impossible. I have subscribed and will watch this development with great interest. Cheers

18
@barnabynorman157 2026-06-18

This is so exciting! An excellent explanation too! Thank you for sharing.

7 1 replies
@LoFiNic2025 2026-06-21

Awesome! Thanks for this, it's getting quite mad what you can do on your own PC these days ..🐇🍄

4
@zzyyxxzzyyxx 2026-06-18

Brilliant video thank you, and exciting project! Open source engineers innovating in a deep way on model architecture, pretty mind blowing

3
@SM26 2026-06-21

First time viewer, thank you YouTube recommendation for leading me here. What an amazing production value. Interesting topic, good script with excellent pacing and delivery. Infographic was clear, and the dashboard with the stats in you demo really makes me envious. Keep it up! I'm here for the ride, subscribing.

1
@adwita7271 2026-06-22

Easy to understand and follow, and very informative. Thank you for the breakdown!

0

Unlock the Data Inside
Turn Videos into Knowledge

  • Get FREE 10/day: transcripts, summaries, chats
  • Chat with videos, export text & PDF
  • $1 free API credit for RAG, chatbots & research

Free forever plan • All features unlocked

App screenshot