Google's Insanely Powerful Response to OpenAI O3! (confirmed)

[@DragonKplus]

Great video again, how did you get access to Flash Thinking? is it any good?

[@Pourya_Kordi_ENG]

Thanks, Yeah it's actually surprisingly good, it's in Google's AI Studio in the model selection drop down inside experimental section.

[@DragonKplus]

@@Pourya_Kordi_ENG cool thx

[@Eriiiiiiiick]

@@Pourya_Kordi_ENG OK_ENG

[@stevo7220]

It is very bad compared to o1 and o1 pro

[@KasperSOlesen]

@@stevo7220 it ought to be... its a flash model, so its the small model, not the full model. It is also why its so crazy fast... and its still experimental.

[@testerdota2]

Google had a truly extraordinary year, when do you remember one company dropping this much technology in just one month?

[@andybaldman]

2003-2005

[@carlosluna6401]

OpenAI during the last three years 😂

[@brianmi40]

OpenAI Shipmas?

[@aiforculture]

This thinking model + NotebookLM's abilities and context window could be incredibly powerful. Would be a lot of fun to upload ten modern books on a subject and ask it to consider what novel ideas are supported or unconsciously infered by all the sources but directly expressed by none of them.

[@abhigavi]

I have a feeling it's going to be AI's of various companies being used to have a big time manipulative game against each other to win and be accepted by more customers and probably later be competition between those AI to rule the world without including any human in it

[@cajampa]

Fingers crossed buddy I hope you are right

[@Saujas]

get off that thing

[@ToolmakerOneNewsletter]

I think its great that you made a video about Google and used so much OpenAI graphics to talk about it. Superb!

[@Pourya_Kordi_ENG]

Glad it helped! :)

[@oker59]

i have no doubt flashthinking is preliminary to Gemini 2.0 - Ultra. Shoot, I think openAI's o3 model is preliminary to what they've got coming in 2025!

[@h.c4898]

There is a beta model called "Gemini 2.0 Experimental thinking" powered by 1206 model. It's currently available in the oaid version. This is pretty good.

[@HoD999x]

where?

[@Zopeee]

You can get it without having the pard version, the 1,5 pro is the only one you need to pay for

[@oluwajuwonloowojori8049]

​@@Zopeeeyou have to pay for it

[@h.c4898]

@@Zopeee in the paid version you have 1.5 Flash, pro (my favorite), Deep Research, 2.0 flash (meh..,) then 2.0 think afvanced (my second favorite)

[@Zopeee]

​@@oluwajuwonloowojori8049not for the 2.0 experimental one

[@calvingrondahl1011]

I hope for the best for everyone.

[@dj0ai]

Great that you are telling the story of the great AI actors. I am exited to hear about Demis Hassabis.

[@sidharthansidhu4783]

Few years later Alpha AGI Alpha AGI-1.1 Alpha AGI-1.2 ...

[@brianmi40]

AGI is a one time milestone. It's reached when the LAST domain (knowledge area) that's been lagging in performance reaches whatever threshold someone finally agrees on is AGI. By definition, all the other domain areas will have already proceeded in large or small ways into super intelligence. So the second AGI is reached, it moves the next second to proceed into super intelligence.

[@picksalot1]

Good video. It is indeed hard to get a meaningful and relevant answer without knowing what question to ask. And knowing what questions to ask depends upon knowing what questions lead to essentially useless answers, and are only a waste of time.

[@فاطمهبختیاری-ف4ب]

Google had a great year for sure thanks for the video I didn't even know flash thinking existed

[@Pourya_Kordi_ENG]

Glad I could help!

[@PencilsAndLight]

Very interesting. Thanks!

[@Pourya_Kordi_ENG]

Glad you liked it!

[@neurojitsu]

Great video, new sub

[@Pourya_Kordi_ENG]

Thanks for the sub!

[@DistortedV12]

don't count google out just yet. o3 is insanely expensive to run. I think Google could make a more efficient model that can "think" on hard mode by end of February & OpenAI seems like they are fighting with microsoft atm. It's just Google is playing catch up which is the problem.ey need to focus.

[@brianmi40]

Deepseek V3 just showed us a lot can be accomplished with a fraction of the training costs. The only true thing is there is no wall or ceiling today.

[@abhigavi]

Google is going to be the boss. After all its a matter of who has more info and data and computing power

[@oluwajuwonloowojori8049]

I think Google is only ahead in terms of computing power. If Claude can get willow

[@jimmysyar889]

@@oluwajuwonloowojori8049 They're google. They have all the internet

[@Gdthainakub]

o3 vs Google AI ☠️🔥🔥

[@Hailzua]

Imagine if quantum computing chips emerge with ai

[@silentz7036]

That’s where I think is a huge problem, if it’s rouge

[@squamish4244]

I still don’t understand how they can function at room temperature.

[@oluwajuwonloowojori8049]

That's what Google is doing. The have the most advanced Quantum Computing Chip (Willow)

[@squamish4244]

@@oluwajuwonloowojori8049 It's all hype though. Unlike classical computing, it doesn't actually mean anything for us at present. We'll either have a superintelligent utopia or a nightmarish dystopia long before it does.

[@anthonymannwexford]

Great video. New sub here. Thank you..

[@Pourya_Kordi_ENG]

Awesome, thank you!

[@oluwajuwonloowojori8049]

In Claude can incorporate Willow. What would we have?

[@Haveuseenmyjetpack]

5:39 this seems wrong to me. A high schooler can simply ask “tell me PhD level information”….

[@user-uk9er5vw4c]

DeepMind is the new King. Goodbye OpenAI. They started researching AI back in the days and now they are harnessing all their researches into business

[@brianmi40]

Yeah, I'm sure OpenAI is shaking in their boots. You know, with all that 12 Days of Shipmas...

[@SimonNgai-d3u]

coz it's just as good as o1 mini but hopefully it will be much more cheaper

[@DBonacich]

After testing all of Google's new year-end models, I can confirm that they are absolutely terrible when it comes to understanding context and especially bad at coding large projects.

[@user-ze9tj9yj4t]

Hi there, thanks for the video, please no music next time - it's not needed, thanks

[@dreamphoenix]

Thank you.

[@mashaallahkordi]

Thanks for the video, OpenAI is too expensive and slow this model seems like a good alternative Google has too much money they can afford free access much more than OpnAI can!

[@Pourya_Kordi_ENG]

Glad you enjoyed the video

[@HoD999x]

why do you say "quote unquote " instead of "quote unquote"?

[@shiftednrifted]

People do that for one or two words quotes usually

[@Pourya_Kordi_ENG]

I'm a programmer we put our quotations in place first :)

[@nyyotam4057]

Yup. This is a terminal race condition.

[@RossPfeiffer]

I hate that I see Sam as the snobby kid

[@NostraDavid2]

The massive amount of vocal fry does not help his case. He's a valley boy.

[@Zopeee]

I rather see him as a deciever, look at his history, the history of OAI in recent times(how they hype stuff up and than it was just tiny bit better than the last models talking about o,o1), or how he wants to change OAI from non profit to for profit, that dude is shady on so many levels its not even funny, i rather think Google has realized that the current path isnt going to lead to AGI, but is rather meant as a tool for the right people(thats why they focus on deving the ai specificaly as tools) and not in the hype way OAI does, since it most likely wont lead to AGI but rather dissapointment(sure it will be impressive, but i dont think it will reach what investors think of it, with the current paths, espically since models like o3 quite clearly show that only sidesteps can make good improvements and not direct steps towards AGI(the results show that it still doesnt have any actual reasoning or understanding, it only mimics those), which wouldnt realy be good for Google so they play the long game since they arent dependent on the hype like OAI is. Even if it turns out that they can reach someform of AGI with the LLm approch(which i find dubious at best, considering it can 'solve'(copy answers) with 'reasoning' PhD level questions, but isnt able to do tasks children can) they still wont lose out that much, so its the safer plan.

[@oluwajuwonloowojori8049]

​@@Zopeee I don't think he's shady in the way you see it. They need a lot of resources to achieve AGI or something close to it

[@brianmi40]

@@Zopeee fan bois of Musk? He ALSO saw the need for a profit entity due to the ballooning costs of training. Facing reality isn't deceptive, and OAI is not the first non-profit to profit conversion, not by a long shot in history.

[@Zopeee]

​@@oluwajuwonloowojori8049maybe it wasnt 100% on Point, but he is most definitivly shady on mutlple levels, again his personal history alone is more than sus.

[@rahi7339]

Their latest model is still crappy

[@muhammadsadri1872]

پرچمت بالاس

[@Pourya_Kordi_ENG]

MerC dadah