Song Han

@songhan_mit

Joined March 2019

250 Following

11.1K Followers

Replies Reposts Media

Song Han
@songhan_mit
Sep 23
Welcome to submit to MLSys 2027. I am thrilled to serve as a PC Co-Chair this year alongside Rashmi Vinayak. We are actively looking for cutting-edge research and industry track papers at the intersection of Machine Learning and Systems. mlsys.org/Conferences/20…
Song Han
@songhan_mit
Sep 17
Not just fast, but iterate fast:
Inco AI
@inco_ai
Sep 17
⚡ 532 tokens/s! DeepSeek V4.1 Flash is live on Inco, and it's the fastest provider on Artificial Analysis. #1 output speed. Not close. Try it: platform.inco.ai Results: artificialanalysis.ai/models/deepsee…
2
Song Han
@songhan_mit
Sep 11
Our Efficient AI course launched in 2022, the same year as ChatGPT, with just 30 students. Then 90, 180, and now 300! Thanks @Shang_mit for co-teaching! All course materials are freely available at efficientml.ai.
Shang Yang
@Shang_mit
Sep 10
Excited to co-instruct MIT 6.5940: TinyML and Efficient AI Computing with @songhan_mit this fall! Looking forward to teaching, learning, and building with an incredible class. Course site: efficientml.ai
3
Song Han
@songhan_mit
Sep 4
Ten years, same question, much harder version of it. Jun-Yan has been pushing the frontier of visual generation — someone I’ve admired since our college days. Now it’s @NunchuxAI. Congratulations!
Jun-Yan Zhu
@junyanz89
Sep 3
Nunchux AI (nunchux.ai) comes out of stealth today. Ten years ago at Berkeley, we spent whole nights on a Titan X in Theano, trying to make iGAN, pix2pix, and CycleGAN real-time so creators could see the results the moment they asked for them. Image models now
Quote
Nunchux AI
@NunchuxAI
Sep 3
Nunchux is live at nunchux.ai! We’re building the frontier of multimodal generative AI inference: fast, affordable, high-quality serving for image, video, and world models. Our Modelverse brings 30+ image, video, and avatar models behind one API. Use
1
Song Han
@songhan_mit
Sep 3
Seven years ago, Muyang started working on this problem before it was popular. Zhekai and Yujun have been building alongside him ever since, and together with Jun-Yan — one of the sharpest people I know in this field — they’ve turned it into @NunchuxAI. Congrats to all of them.
Muyang Li
@lmxyy1999
Sep 3
Today is a meaningful day for me. Seven years ago, as an undergraduate, I started researching efficient generative AI. At the time, few people saw generative AI becoming something people would use every day. GANs were still the frontier. Then came diffusion. Now we are beginning
Quote
Nunchux AI
@NunchuxAI
Sep 3
Nunchux is live at nunchux.ai! We’re building the frontier of multimodal generative AI inference: fast, affordable, high-quality serving for image, video, and world models. Our Modelverse brings 30+ image, video, and avatar models behind one API. Use
1