Am ajuns efectiv la scenariul din Her.
Discut cu claude opus (e drept technical stuff) in timp ce stau cu copii la joaca.
Design ,architecture,options ,risks.
All voice..
Mai foloseste cineva vocea?.
Știu ca se poate si in claude code dar si in Copilot dar se obicei sunt la tastatura...
Practic pentru unii este sfarsitul ruby on rails si al programarii.
Pentru altii, un nou inceput
Fascinant cum prezinta.
TLDW : Nu mai scriem cod dar Jevons paradaox -> totul o sa fie bine.
Niste leaked benchmarks aratau scoruri incredible .. o sa fie interesant.
Din septembrie am zis sa am si eu bugetul organizat cu categorii, cat dau pe fiecare categorie, economii, facturi si tot ce trebuie.
Nu stiu cum va iese voua dar eu credeam ca ma incadrez cam in 10-11k ron pentru familie de 3 dar punand bani deoparte pentru orice (fond de electrocasnice, fond de telefon, fond de laptop, fond de renovari, fond de reparatii auto,fond de analize/consulturi ) ajung la 15k RON/pe luna lejer.
Daca iau doar ce arde (intretinere + facturi+ mancare) ies mult mai jos pe la 4-5k RON.
E mult mai greu pentru ca am renuntat la cardurile de credit si atunci trebuie sa ma pregatesc in avans pentru orice ( moare masina de spalat, trebuie sa am 1000 RON pregatita)
Concediul nu mai merge sa il las in "aer" si sa il platesc in rate .. (aveam cardul ING doar pentru asta). - trebuie sa pun bani deoparte lunar.
Voi cum faceti cu bugetul ?
Pe scurt.
Am avut 3 carduri de credit
1)ING
2)Starcard bt
3)RZB emag.
Le-am folosit pe toate mai ales la cumparat in rate fara dobanda .
Totusi planul meu era sa folosesc cashback,stoozing si alte tricks prin care sa si câștig macar comisioanele anuale
Din păcate cu exceptia cardului rzb care aparent mi-a adus 4xx puncte (adica lei),restul au adua beneficii minime.
Parca 100 ron pe an la starcard si zero la ing...
Am inchis toate cardurile,mai ales ca si planurile de rate mi se pareau gresite ...
Alternativa mea e sa creez fonduri de economii tintite pe achizițiile viitoare.
Efectiv nu vad niciun avantaj.
In sua mai iti dau airline miles,cashback de pana la 5%,asigurare ...
In europa nu vad avantaje (in afara de primit linia de credit din spate)
Imi dau seama ca cele mai mari crize din istoria bursiera s-au intamplat toamna :
https://en.wikipedia.org/wiki/Black_Monday_(1987)
http://en.wikipedia.org/wiki/Great_Recession
si bineinteles:
https://en.wikipedia.org/wiki/Wall_Street_crash_of_1929
Sell in May and go away dar eu sunt inca cu index investing si no market timing.
Ce parere aveti?
Unde e saptamana aia de lucru de 4 zile? Probabil in profitul actionarilor ...
salut
Azi am platit toate cardurile de credit (3) si vreau sa ma apuc de economisit pentru vacante si electronice .Erau pline de rate fara dobanda de dadeam 2000ron pe luna totalul..(le inchid săptămâna viitoare).
Gemini zice sa folosesc revolut savings ca sa am conturi separate pentru fiecare obiectiv.
Am incercat si anterior cu ing dar nu prea am reușit (tot mutam banii inapoi in contul principal).
Voi cum faceți?
Shocking dear ...
Nu e niciun suspans, index investor deci pe termen scurt nu fac nimic..
Totusi ma aștept la un mini crash luni dimineață dacă (sau mai exact când) vine rezultatul raportului
Cu Fitch au mai negociat, cu Moody's nu știu dacă merge...
Am și eu o curiozitate...
S-a întors cineva la birou după 6 ani sau mai mult de lucrat remote?
Eu sincer nu cred ca as mai putea, adică cât am lucrat în open space abia dacă livram ceva în timpul programului. Făceam over time ca să lucrez în liniște..
Ma tot contactează recruiteri și deja au băgat-o pe aia cu 4 zile hibrid... Bună gluma. Mai bine ies la pensie..
Putin critical thinking nu strica
https://www.kimi.com/blog/kimi-k3
Benchmarxxing sau chiar e asa de bun ?
De ce as cumpăra chestia asta?
Openai a luat-o pe ulei...
Pare ca vom mai avea 7 zile.
A doua extensie după 8 iulie.
Csf ncsf
Mulți boti rusi și chinezi care ne zic ca ne ia AI apa și energia (mai lipsea hrana).
Grok 4.5 was released and they claim they offer a performance similar to Opus 4.7 with half of the cost. While this remains to be seen and to be confirmed, let's see how this claim seems to follow a trend we are experiencing the last years.
I have collected all the data of the Epoch AI, which is combination of various state-of-the-art benchmarks (e.g., MMLU, GPQA, coding, reasoning tests), combined into 1 score, the ECI (Estimated Capability Index).
The ECI score renders reliably in between two models, from model A to model B, meaning it can capture the difference of the generic capabilities between two AI models.
In March 14, 2023, the GPT-4 model was released, and it has an ECI score of 126 points.
Scenario: Let's suppose now that someone wants to achieve something that is sufficient to be achieved with an ECI score of 126, so they want to "get a level of intelligence of ECI 126 by paying the absolute minimum cost currently in the market".
During that time (March 14, 2023), to get the level of intelligence of ECI 126, you had to pay $37.5 (input/output blended). Today you have to pay $0.13.
This is a 99.6% decrease in price for the exact same minimum level of intelligence.
Here it is a graph to understand the drop of the cost.
Cost of getting a level of intelligence of at least ECI 126 over time
In January 20, 2025, DeepSeek-R1 became the first model to hit the ECI 140 mark. At that time, it set the initial minimum cost for this intelligence tier at $0.96 (input/output blended).
Within just three months, the price floor collapsed with the release of Grok-3 mini in April 9, 2025 brought the cost down to just $0.26 while maintaining an ECI 141. This is a 72% decrease in cost over a very short period (3 months).
Cost of getting a level of intelligence of at least ECI 140 over time
Continuing, the GPT-5.1 was released in Nov 13, 2025 with an ECI 150 and with a cost of $3.43 (input/output blended) to achieve it.
Four months later, on December 17, 2025, Google released Gemini 3 Flash with an ECI 151. This release shattered the previous price floor, bringing the minimum cost down to $1.12 (input/output blended) . This represents a 67% reduction in cost in just over four months.
Cost of getting a level of intelligence of at least ECI 150 over time
Ultimately, data shows that as a general rule for the current market, you can expect:
The price of an "X" level of intelligence to drop by at least half, every 2 to 4 months.
This is absolutely incredible to think about it. We don't know what exactly ECI score could somehow represent a "human level intelligence", but given that human level of intelligence is fixed over time and it is not moving, if we continue like this there will be a point in time that we will reach "human-level intelligence" with a cost that is nearly nothing to consider about.
anyone else coming to terms with this?
I'm trying my hardest to come up with prompts to throw at Fable that it will actually get wrong, and I can't find any.
I have one more big one for tomorrow when my limits reset. I have a game in Unity. It's filled with custom plugins, spaghetti code, and literally every part of it is a mess. I am going to tell Fable exactly one thing: "Port this game to Godot. Make it functionally the same" and see if it can do it.
If it can, I think I'm gonna start thinking about what jobs are safe from AI.
Bate Fable se zvonește..
