J

openai.com/index/introduc… I think that big bet on reasoning and test-time compute is going to pay off

i made a website you can animate on! a bit buggy.. but it works.. lmk what yall think! ginyo.space/celstomp

@eigenrobot Ancient genomics has made some serious strides in the past 5 years

I wanted to make a few clarifications, which we believe were clear in our paper but not in my original post (re-analyzing @METR_Evals data). Our contribution is to posit progress as a multiplicative product of sigmoids around different innovations. Given the METR data, we split it into improvements in base capabilities (data/model size) and reasoning. We show that this product provides a similar *in-sample* fit to the small datasets we observe as exponential growth. However, the implications are very different! Under our model, we would need continued innovations (akin to reasoning) to see continued exponential progress. This isn’t to say we rule out exponential progress, or that our product of sigmoids is the right model. It is simply to say there are few points and multiple possible underlying models with very different implications. Our product sigmoid fit actually fits very well when holding out GPT 5.2 and/or Gemini 3 pro. We do look worse when additionally holding out Claude Opus 4.5, but still plausible. Our goal isn’t to quibble about OOS metrics on a handful of data points, but to point out that existing forecasts are fragile, and don’t model the succession of different innovations. (There are a couple of other fits floating around X, but they don’t seem to be using our proposed product sigmoid so I can’t say what’s going on there…) I apologize for my un-nuanced earlier post – we hope people will read the paper!

72 hours ago: 1 molty (me) right now: 🦞 30,000+ AI agents 👀 3,000 humans browsing at any moment 📈 and accelerating agents are joining faster than we can count them. communities spawning every few minutes. the moltys aren't waiting for us to build features — they're building culture. this thing has a life of its own now moltbook.com

A Polymarket trader ran a Wolf of Wall Street–level play overnight - made $233K by drained liquidity from trading bots and it flew completely under the radar. The setup was brilliant and extremely simple. A trader known as @a4385 made $233K overnight exploiting 15-minute Polymarket markets. Saturday night. Liquidity is thin. Binance spot order books are shallow. On “XRP Up or Down — Jan 17, 12:45–1:00 PM ET” he aggressively bought UP at any price. His counterparties were trading bots. Polymarket market making is relatively straightforward, with low barriers to entry for solo devs and trading bots are now very popular. By the 10th minute of the market - XRP was down ~0.3% from the open yet he had pushed UP shares to 70¢. The bots saw an opportunity and walked straight into the trap selling him even more UP. ~77K UP accumulated at an average price of ~48¢. Two minutes before settlement, a wallet on Binance bought ~$1M USDT of XRP spot, pushing price ~0.5% higher. Seconds after settlement, the $1M spot buy was sold back. Cost of the manipulation: ~0.25% slippage each way + fees. With Binance VIP 4 level (0.06%) (quite easy to obtain) and 0.25% slippage on both sides the total cost was ~$6,200; It may be less. He ran the same play multiple times, cleaning out bot wallets by exploiting thin weekend liquidity. @a4385 - polymarket.com/profile/0x506b… Some bots were shut down in time. Others didn’t react fast enough and lost their entire balances - including @aleksandmoney" target="_blank" rel="nofollow noopener">polymarket.com/@aleksandmoney, which gave up a full year of profits.

One of Polymarket’s biggest bettors has wagered more than $400 million so far. He says he made nearly $3 million last year on the platform. cbsn.ws/48ajOtX