So Me

@haider1 Cope

google has already secured its position they don't have the same urgency to win immediately with gemini being integrated across their services, they can afford to wait, focus on research, and train stronger models the last six months focused more on harnesses than research

@GoblinSPL Stalling

GPT-5.6 waiting room🧌

@synthwavedd come close 5.6 come close

iris-alpha

@l_mejiaC @jxnlco /meaning

@jxnlco /purpose

should make an ultragoal skill huh.

@thsottiaux Whatever field you in,whatever excites you,the best place you could be in right now is in a frontier AI lab.

DM me if you have a knack for marketing to a demanding crowd

@haider1 @polynoamial hook me up with that hoodie big dawg

OpenAI Noam Brown says single-number benchmarks no longer make sense for modern AI models Once models can use CoT and extra inference compute, their performance depends heavily on how much reasoning time they are given "it made sense for GPT-2/3/4, but not for reasoning models"

@l_mejiaC 2028 and beyond.don’t die

I can’t wait for daily and hourly releases. Even releases per second.

@LexnLin It’s a fight to the finish now between OAI and Anthropic(Google Proxy).Enjoy the show

This Google I/O was pretty disappointing. I expected wayyy more. At least a sota model not just a flash 3.5 thats even really expensive Basically what is cool is the concept behind the omni model and the redesign, but otherwise nothing surprised me and seems to be new. Antigravity app UI looks like Codex app UI Antigravity CLI looks like Claude Code Subagents, building android apps, design and so on was all possible before. I've been doing all that in Codex or Cursor/Windsruf and it's perfect. The limits are also pretty bad now. Good for just using 3.1 pro, 3.5 flash. BUT on Ultra you just don't have the value anymore. Before you could generate 5 videos and still have everything else seperately. Now 5videos in and I can't use ANYTHING anymore. Antigravity seems to be still broken with the errors. The tools in Flow are clearly vibecoded (and I'm not saying that this is bad). It just doesn't look good. A LOT of new features like Spark for example are only available for Ultra users in the US. I have been an Ultra subscriber since 3months. Not getting Genie 3, Spark, Agent, still no personal intelligence,... (ah yes and apparently Genie 3 is available for all Ultra subscribers now but doesnt work for me) This is not worth it anymore. I don't see the value.

@LexnLin lol

I can't even imagine how fat the cheque was

should i marry chatgpt when the new voice update comes out

Personal update: I've joined Anthropic. I think the next few years at the frontier of LLMs will be especially formative. I am very excited to join the team here and get back to R&D. I remain deeply passionate about education and plan to resume my work on it in time.

@annapanart Week after

GPT 5.6 next week?

@haider1 Late May Enjoy

if openai keeps the same release pattern from gpt-5 to gpt-5.5, around 50 days gpt-5.6 should launch around early june but they might speed it up, especially if google launches something like gemini 3.2 or 3.5 at i/o that would be a good moment for openai to respond

For those of you living inside the codex app, what should we prioritize among features, reliability or performance?

@thsottiaux @argofowl

@argofowl Hmmmmmm

haven't used codex that much yesterday which means i have a lot of usage in reserve which means i can smash 5.5 xhigh on /fast for a while @thsottiaux don't you dare reset my limit today sus boy i'm keeping an eye on you

@JinjingLiang Need ultrfast xxxhigh fat chungus in times like these

🚨 Codex quoats will reset "this evening" Turn on /fast mode, use it all!!

@yacineMTB Goalify your prompts

Never use a codex prompt without /goal by the way

@thsottiaux @argofowl What was the something fun you released yesterday? Missed it

@scaling01 Vapor

It's over Mythos is insane