Aidan Clark
2,010 posts
Aidan Clark
@_aidan_clark_
Joined November 2020
o3-mini first try no edits, took 20 sec (told me how to convert to gif too.....) Get excited :)
GIFThis post is unavailable.- Wow Sonnet 4 seems amazing, congrats Demis & Team!
- 2x cheaper & faster is for English, but for other languages (especially non-latin-script) expect - thanks to our new tokenizer -- even up to 9x cheaper/faster!1.7x fewer tokens in Korean, which means GPT-4o feels 3.4x faster to Korean users!
- gpt-oss is our new open-weight model family! the bigger one runs on a single GPU, you can run the small one on your laptop. Go install it right now, seriously! Telling your laptop to do something and watching it happen made me feel the AGI like nothing since ChatGPT.Our open models are here. Both of them. openai.com/open-models
- Hi, We’re delaying the open weights model. Capability wise, we think the model is phenomenal — but our bar for an open source model is high and we think we need some more time to make sure we’re releasing a model we’re proud of along every axis. This one can’t be deprecated!we planned to launch our open-weight model next week. we are delaying it; we need time to run additional safety tests and review high-risk areas. we are not yet sure how long it will take us. while we trust the community will build great things with this model, once weights are
- o3-mini's intelligence x speed combo is incredible, idk what to say other than just try it and see for yourself. This took 8 seconds, how long would it take you?
00:00 - Okay I’ve had enough extremism: I’m founding an AI Centrist Party. Tenets: * exponentially improving AI isn’t right around the corner * LLMs are a massive step in AI capability for any good definition of that word * worrying about AI risk is reasonable * retweeting Yud is not
- i love the openai team so much
- Asking a tokenized LLM to count letters is like asking a colorblind person to distinguish aliased colors: sure, it could use its intelligence to deduce color in clever ways but it’s a task which is fundamentally harder for it for uninteresting reasons.
- On Tuesdays we usually swap in GPT5 for the plus tier but on Thursdays some people get the initial version of 3.5T with the bug in it, really it’s anyone’s game gotta keep people on their toes.actually annoyed by this. due to randomness and confirmation bias people always try to claim chatgpt changed when it hasn't. but now they are actually updating it without telling anyone, so these speculations will never end.
- I got disillusioned with RL when I realized that it was always: step 1: act randomly for ~years worth of data before stumbling upon a reward step 2: figure out how to repeat that action in a generalizable way .... and no one had good ideas for improving step 1
- @ren_hongyu killed it To recap the demo (I'm still sweating), o3-mini wrote it's own ChatGPT UI to talk to *itself* via the OpenAI API, we asked o3-mini to write and execute a script in this UI to evaluate *itself* on GPQA, and the resulting script correctly returned 61%.
- Dividing research from engineering is so weird. Good engineering is (systems) research and so many great engineers have the same traits that I see in great researchers. IMO the real thing is: SOTA AI now depends on good systems knowledge/innovation as much as good ML knowledge.
