Sabitlenmiş Tweet
@GeostatsGuy @X I will share this and pin on my profile, solid advice!
English
Dr. Emil Mirzayev
3.1K posts
Dr. Emil Mirzayev
@MirzayevEmil
🇦🇿 Senior Researcher at @UCLSoM; **I am on 2026/2027 academic job market** https://t.co/rnhgBapOVx I write about AI and strategy. like!=endorsement;
London, England Katılım Mart 2017
213 Takip Edilen389 Takipçiler
@zaurs Ola bilər ki, bəzi şeylərdə bəxtin gətirməyib, amma Güneldə cekpot udmusan😀
Türkçe
Bizdə dəyişib. Yəqin bütün ölkələrdə dəyişməyib.
JADKSNDKSNXKSNXKSNCSLXNXNSKXNSKXNSMX@itsmesalesta
@bpthaber Bende hala aynı yalan haber yapmaktan ciğeriniz çıkacak ak
Türkçe
@evfcfaddict @Picolinie They paid like 40x ebitda for them. That's too much imo
English
@Picolinie Massive medtech sector weakness + market seems to dislike their recent m&a deal (Exact Sciences) as well.
English
@emollick It used to be quite a powerful move. In our SMJ paper (2025) it proved to be a good way to scale AI evaluators horizontally to achieve wisdom of AI crowds in strategy evaluation.
English
A reminder that telling the AI that it is an expert in a field is no longer helpful in making the AI better at that field.
Ethan Mollick@emollick
We tested one of the most common prompting techniques: giving the AI a persona to make it more accurate We found that telling the AI "you are a great physicist" doesn't make it significantly more accurate at answering physics questions, nor does "you are a lawyer" make it worse.
English
@TidefallCapital Is there a good free write up to get started learning about the business?
English
$SFM Q1’26: sales +4% YoY, comps -1.7%, EPS $1.71, 6 new stores (483t), $252M cash, 0 revolver debt, $140M buybacks, ROIC 17.4% vs 16.8% LY. Growth drivers: loyalty adoption, >1.5K new items, near-complete meat self-dist., strong new-store productivity & supply chain optimization
English
English
@VB4alpha @Etnarosso1973 fcf will massively ramp in 2-3 yrs
Intred has lower churn and is more predictable and better managed
English
Intred $ITD.MI AGM gave authorization to buy back up to 1,5 million shares in the next 18 months (33% of the total free float).
Keep on selling this 46% ebitda, low churn business at a 17% operating cashflow yield….
cms-admin.intred.it/uploads/PR_Int…
English
@zaurs Şah plovun pr-ı yaxşıdı. Ləvəngi plov və ya fisincan plov ən yaxşısıdı.
Türkçe
Şəkidə plovun bəlkə 50 növü var, amma şah plov nəmənədirsə, mən onun adını son 10-15 ildə eşitmişəm. Şadlıq evlərinin uydurduğu bir şey kimi gəlir.
Ramin SADIK@saki552003
Bağımsız Türk ülkeleri mitfaklarının en ünlü yemekleri Hangisini tercih ettiniz, yahut ederdiniz?
Türkçe
@soumitrashukla9 They nerfed the thinking time on 5.4 pro (normal and extended thinking) by about 15x. Things that required about 30-40 min thinking now finish in 2-3 minutes. Quality is also way worse
English
Anthropic: Keeps limiting compute and lying to playing customers / nerfing models.
OpenAI:
- 10 min downtime? Limits reset!
- We hit 4 million followers? Limit reset!
- Starbucks person spelled my name correctly on my coffee today, let’s have a limit reset!
Tibo@thsottiaux
Happy Tuesday. Codex has hit 4M active users, adding over 1M users in less than two weeks. To celebrate we will reset the rate limits again in a few hours. Enjoy!
English
@pedrohcgs @Apoorva__Lal @soumitrashukla9 @alexolegimas Seems like openai has needed 5.4 pro a lot. Thinking on pro reasoning (which is available only to pro or ultra subscriptions) lasts 10x shorter.
English
For all my AI/Codex/ChatGPT people here:
Can I use ChatGPT 5.4 Pro inside Vscode? Can I integrate it with Claude Code, in VS Code?
Is anyone using it like this?
Any pointers on how to integrate?
Tagging for reach: @Apoorva__Lal @soumitrashukla9 @alexolegimas
English
@kimmonismus They have even reduced the reasoning (nerfed by 10x at least) of 5.4 Pro reasoning mode (ultra and pro sub product). It does not seem much different from Antrhopic in that term..
English
@TheRealAdamG and chat stops (nothing happens, no reply or whatsoever) immediately when I say I want to talk to a human representative.
English
@TheRealAdamG Thanks, Adam. The support bot appeared to be repeating the same thing (it can be UI issue, not selecting pro model explicitly, sometimes it can answer fast). I have tested it myself with about 20 different prompts (w var. difficulty) and reasoning was 1-3 minutes in each.
English
openai.com/index/introduc…
I think that big bet on reasoning and test-time compute is going to pay off
English
@TheRealAdamG I have reached out to them. Waiting reply. In the meantime, I checked on reddit and OpenAI support forums to see that was the case. So, it is not one time off thing it seems. Example: community.openai.com/t/chatgpt-5-4-…
or reddit: reddit.com/r/ChatGPTPro/c…
English
Dr. Emil Mirzayev retweetledi
@evfcfaddict got hacked and is having trouble getting help from @X.
Be extra careful about which links you click on. #helpandy
English
@nmboffi For me it went from like 30-40 minutes to 2-3 minutes of thinking. More than 10x reduction.
English
did openai nerf 5.4 pro? it used to churn for like an hour on real mathematical research questions and i used it extensively. now it outputs a generic response after 5-10m that is significantly lower quality and much less useful
English
User pays pro price, but gets nano model reasoning. Nice business model imo :)
English
Seems like OpenAI has reduced the reasoning duration (ability) of GPT 5.4 Pro at least 10-15x (from 30-40 min to 2-3 min on my tests). That is as big as a rugpull as Anthropic has done with 4.7 "adaptive" (read: minimal) thinking.
#openai #chatgpt #anthropic #reasoning
English