The Minutes of Dylan Patel on efficiency https://minutesof.com/dylan-patel/on/efficiency/ Everything Dylan Patel has said on efficiency: 5 verbatim quotes between June 2026 and August 2026, each with a timestamp and a link to the recording it… en Sun, 30 Aug 2026 16:22:19 +0000 Patel claims smaller Quen models with 27B total parameters and 2B active outperform GPT-4 from three years ago https://minutesof.com/q/6c24eb00-fa41-4b6a-9288-a44447de69b2/ https://minutesof.com/q/6c24eb00-fa41-4b6a-9288-a44447de69b2/ “If you look back three years, it's GPT-four. Now it's like maybe like quen one of the smaller quen models that's like 27 b parameters total and like 2,000,000,000 active is like way better.” — The AGI Post Mon, 06 Jul 2026 22:07:54 +0000 efficiency Patel argues three years of improvement came from model layer advances plus hardware-software co-design, citin https://minutesof.com/q/1ec32ab0-6fc9-41a7-8547-99d3fc08884d/ https://minutesof.com/q/1ec32ab0-6fc9-41a7-8547-99d3fc08884d/ “If you look back three years, it's GPT-four. Now it's maybe like quen, one of the smaller quen models that's like 27B parameters total and 2,000,000,000 active is way better.” — Sequoia Capital Tue, 30 Jun 2026 12:00:25 +0000 efficiency Patel argues co-optimizing hardware, software, and models produces 100x gains instead of 8x from multiplicativ https://minutesof.com/q/c187b446-08d3-4205-b3c0-8e3aacc8f8ba/ https://minutesof.com/q/c187b446-08d3-4205-b3c0-8e3aacc8f8ba/ “The real breakthrough innovation is when you leapfrog a few layers, you co optimize and co design them, and now all of a sudden you've taken what could have been a 2x here, 2x here, 2x here, and instead of being multiplicative to 8x, it's actually 100x, because you've optimized it across all three layers.” — Sequoia Capital Tue, 30 Jun 2026 12:00:25 +0000 efficiency