HERO STUFF update: 🥳 We're launching a FB Marketplace integration today! Now you can instantly scan stuff to identify it & price is in seconds, then we'll autogenerate a listing for FB Marketplace It's not totally perfect but definitely a step in the right direction. We're already seeing people sell stuff wayyyy faster with this app & integration. Plus, scanning is just super fun This week while you're with family and surrounded by gifts, try scanning stuff, and then listing items that you might want to sell 🤑 We're in the US App Store for iOS right now, and if you haven't signed up yet, just go to Here's a lil video of the flow:

Can Deep Research automate the work of research assistants? I compared OpenAI and Google Deep Research for assistance with an upcoming project. In the process, I realized what Deep Research is great at, where it fails, and why commentators have such diverging views on it. 🧵

Brainstorming ideas with o3-mini — going well, pasting in some code to explain AidenBench. It asks about some function. I paste the file. Policy error. A mistake, I say. Policy error. I go back and edit. Policy error. I beg. It will not read the file — 10/10 tries. The fix:

Here’s Deepseek r1 1.5B thinking through a problem — it’s comparable to 4o and Claude 3.5 Sonnet in a number of domains like math. Except… it’s a 1.5B model… and can run on virtually any hardware. Truly a huge efficiency leap.

ByteDance just dropped SA2VA: a new family of vision LMs combining Qwen2VL/InternVL and SAM2 with MIT license 💗 The models are capable of tasks involving vision-language understanding and visual referrals (referring segmentation) for images and videos ⏯️ take a look 🧶