ByteDance is will launch Omni-human model on jimeng (English called dreamina), and it's a closed source model.

Report: Ilya Sutskever's startup in talks to fundraise at roughly $20B valuation

Meta torrented over 100 terabytes of pirated books from Anna's Archive, Z-Library and LibGen for AI training

Imagine if this happens this year

There are now at least 3 truly telepathic people thanks to Neuralink

New Gemini 2 Pro scores WORSE than the old version at most categories on LiveBench

Gemini 2.0...um, this is a wall

Gemini 2.0 (Pro, Flash, Flash-Lite) Now Live in Arena: Pro Claims #1, Flash Available to Devs at $0.1/M Tokens

Google officially releases Gemini 2 Pro and two new versions of Gemini 2 Flash

Gemini 2.0 Pro has similar benchmark scores to Sonnet

OpenAI Sales Associate Agent spotted in recent Tokyo talk (zoom in, pic is from livestream)

Today, I made the decision to leave our Collaboration Agreement with OpenAI. Figure made a major breakthrough on fully end-to-end robot AI, built entirely in-house

Universal Jailbreak Testing is up for Claude

Is This The End of Non Fiction Writing?

Just imagine.

Next OpenAI event in two days. Most definitely o3-mini related, but what could it be?

Sam Altman: "Every year we move one standart deviation of IQ."

Huge news: Open Ai is developing hardware which will be more disruptive than the Iphone

Anthropic has better models than OpenAI (o3) and probably has for many months now but they're scared to release them

All Benchmark Results of Deep Research

OpenAI employee - "Using Deep Research over the past few weeks has been my own personal AGI moment. It takes 10 mins to generate accurate and thorough competitive and market research (with sources) that previously used to take me at least 3 hours."

We're sooo back Gemini 2.0 Pro Thinking is imminent

AI researcher discovers two instances of R1 speaking to each other in a language of symbols

The "stop competing and start assisting" clause of OpenAI's charter could plausibly be triggered any time now

You only get 6 o3-mini requests every 3 hours for the free version (which scored lower than deepseek-r1 on livebench). Funny, I thought they were going to release o3-mini for everyone for free unlimited uses, and o3 for paid users. Guess once you run out of 6 free requests you go back to 2023.