@sama has the most realistic and optimistic AI perspective of anyone. @OpenAI is the one to lead it. Demis never talks about unemployment and only about positive aspects of AI. Dario mainly the negative effects. Thank you @sama for releasing GPT and keep giving us better models. And bigger too please. GPT 5.6 Ultra :) Don't slow down, accerlerate faster. The inevitable pain of transitioning for some will be shorter that way. You've been the main push to change the world into AI and will go into history.. many appreciate that 🙏

OE is very mid, has always been very mid. pay 20/100$ for GPT-5.5 and set up some custom instructions. it's way superior. supplement with Opus 4.8 ( Gemini 3.1 pro) you can also try the free HIPAA x.com/thekaransinghal/status…

This was totally anticipated by anyone who uses paid AI openevidence has always been mid since 2023

and this is based on GPT-5.2 not 5.5 soon 5.6. I have always kept saying, that openEvidence is very mid.. good idea, but subpar to GPT/Gemini/Claude. I still don't get why so many MD/DO/PA/NPs use it, I assume good marketing? DoximityGPT is better and GPT for clinicians is free too now and much better and HIPAA too

Gemini 3.5 extended and 3.1 pro extended get it right with this prompt “Solo 3 parole: non sei solo” into english & german in local way of saying it with translation to get the accurate saying

epoch.ai/benchmarks/simple-q… this seems to be related to model size, except for qwen3max, but 5.5 seems larger than opus, smaller than gemini

So anthropic wants 30d mass data surveillance on all its customers? While I like opus 4.7/4.8, fable is not usable. Hope OpenAI wins out of all and people rethink what anthropic is trying to do..

The OAI / Anthropic values difference is deeply misunderstood, even within the walls of both. Should a loving ensouled machine God watch over humanity? Vote Anthropic. Should humanity be entrusted with the tools of its own progress and destiny? Vote OpenAI.

i mean it seems we will never get AGI if fable already freaks Dario out. True AGI will make Fable look like low-IQ.. it's all overblown. it's not like someone is going to make Covid 2027 in their home lab tomorrow.. not a single medicine questions works..

does openAI need a bigger model like fable? and give it for Prox20/200$ payers? Sure that would be cheaper than serving PRO models, at least based on API?

I was hoping Fable would be a killer and work even longer than Opus 4.7/4.8 and then upgrade to Max200 plan.. but per the blog post: "Therefore, for the time being we have arranged for Fable to fall back to Opus 4.8 on most requests related to biology and chemistry. As with all of our classifiers, we hope to narrow these safeguards as soon as possible: as can be seen from the evidence above, there is great potential for positive applications of Fable for science, and we do not want false positives from our classifiers to get in the way" It won't even search medical trials lol

Even gemini 3.1 / 3.5 is more usable than Faboe 5 ; outside of coding.

How did it score that good on healthbench when it doesn't answer medical questions for me..

I am sure soon (1-3y) AI will be smarter than 99.9% of all humans, people should learn to accept and deal with that and adjust. Otherwise they'll have pain.. Question to figure out with time, past the hype, will it even be smarter than near all humans (99.999%) or even all humans including Einstein and Nobel prize winners? We will see.

I'd say some jobs out there are overpaid, significantly. very few humans are worth that much, and most in those jobs are not them..

youtube.com/watch?v=TjrShuj_… 24:20: the next big training run in the fall on Vera Rubin >> GPT-6 end of 2026/early 2027 ?!

youtube.com/watch?v=TjrShuj_… very interestingly around 20m:45 she describes 5.4 as a "really large pre-trained model" that was significantly more expensive and harder to serve than their newer releases (assume 5.5). so was 5.2/5.3/5.4 a new pretrain and now larger than 5.5?? 5.4 was awfully slow compared to 5.5

If we can use Codex also in the cloud when the laptop is off (and sync back when online) that would be amazing. That would fully replace the old o3 Agent.. It would be great then to have several daily automations for me, e.g., digging journals for new literature in my specialties and summarizing for like a daily 3-5k word summary of 20-50 studies :)