Introducing Gemini 2.0 | Our most capable AI model yet
Вставка
- Опубліковано 5 лют 2025
- We're announcing Gemini 2.0, our most capable AI model yet that’s built for the agentic era. Gemini 2.0 brings enhanced performance, more multimodality, and new native tool use.
Learn more about Gemini: deepmind.googl...
Read our Gemini 2.0 launch blog post: goo.gle/4g2b9Lw
Subscribe to our Channel: / google
Find us on X: / google
Watch us on TikTok: / google
Follow us on Instagram: / google
Join us on Facebook: / google
Gemini worked great for me but now I'm so excited to update it with 2.0, my stack of CrewAI, Composio and Gemini 2.0 is gonna be overpowered!
I don't think they can topple the almighty OpenAI
Isn't the Google the almighty?
OpenAI for the win
the war of the ai (models)
Loved your stack but would replace crewai with langchain
What a time to be alive
We were truly born in a good era. We fully understood the past and the future.
Grab your papers
❤❤❤❤🎉😢
bro, i am so funking done. ON EVERY AI VIDEO "what a time to be alive!" ARE YOU A Ŧยςкเภﻮ BOT? I HATE SEING THIS COMMENT IT SOUNDS LIKE YOUR A BOT.
@@999satyam you understand it
presentation looks like one of thr collectable videos you recover in a post apocalypse game and how it all started.
like a fallout audio tape found in some decrepit old facility in the wasteland haha
I WAS JUST THINKING THAT
Haha so true!!! reminds me of horizon zero dawn!
10/10
hahahahahahahah🤣🤣🤣🤣
The AI race is wild; I can't even imagine what the tech world would look like in 5 years.
@@amarug 5 years is a very very long ways away with where our technology is already at, even with incremental progress we still are close to AI capable of being able to creatively start to build new advanced tech by itself. Some already can, like the AI that designs chips by itself and makes them with virtually zero human intervention. AI is capable of designing most hardware better than almost all humans can now. If even narrow ASI in programming or math or one field happens in the next few years, it'll completely crush software development or challenge all our known theories and with almost zero to no human intervention too. All it could take is one or two breakthroughs and a couple years to sky rocket progress (we already have AI researchers saving humans thousands of hours), and not just steps or leaps but exponential evolution. AI progress will get faster the more advanced we get and more embodied it gets in the world. Having potentially hundreds of thousands or millions of humanoids combined with LLMs and several other technologies by 2030 will make a HUGE difference. Because they can to some extent learn autonomously already and will have a combined millions of years of experience in the real world (or simulated words that are generated at the same degree as our real world since simulations are getting better) (also can combine each experience and they update via some network and learn from each others experiences), they could evolve faster than humans at a certain point. One instance of this is the robot dog that AI programmed it (not a human) to learn how to balance on a ball while humans kicked it, and did so in a virtual world first and then the physics somehow worked in the real world!? If that kind of thing is already possible, it's hard to imagine how much better things will get in even just a few years from now! 5 feels like 50 years to me.
Jumps ended before gpt4, jumps ended before 4o, jumps ended before o1, jumps ended before avm and sora, and now gemini 2.0
What about gpt4.5 and gpt5? What about agents with low latency and low cost? What about doing youtube videos with ai video? What about o2-3-4-5? What about project Astra? What about claude 4.0?
You don't know nothing please go bearish they will destroy your money bro@@amarug
Think when these models start to utilize quantum computing. Skynet starts to look like nothing compared to this.
imagine going to prison in 1995 and coming back out in 2030 💀
@@depralexcrimson you be like a newborn, entering a new world again😅
Oh hell yeah, we're finally getting the real Google Assistant. 🌠✨
The full, tailored experience will likely be exclusive to the Pixel. Still, I hope it'll improve for the rest.
@@baldierot Pixie
@@baldierot It'll be exclusive to pixel if they want to lose a huge market, sure. They're going to develop a game assistant AI and make it a smartphone exclusive? Skipping every console and PC player on Earth?
I guess they might. Will see.
for the first year its will only avaliable in pixel, the next year, then it will be in all android
Thing is google never actually releases anything more than demos for anything good..
oh google, couldn´t wait until openais week is finished?
Of course NO! Why will they let them have all the fun when they have something like this lying around???? 😂😂
All Google does is announce.
thats what openai did to them last time google announced lol
Damn I already spent $200 on pro
@v1nigra3 the difference between open AI and Google is that the announcements that open AI is there any right now become available the same day
1:17 Tamil language featured ❤️🔥
itha pathi vera ethavathu solla mudiyuma?
gemini 2.0: alphabet ceo sundai pichai, ithuku mela sollava venum?
World first language ha marakka mudiyuma❤
the noodle language
@ MixbOosted It is better than your russian font and we have meaning for every curves and bends in our language. Does your language have meaning for your language letters bend. Mind your words before talking about other languages.
Tamil / தமிழ் ❤ The world’s oldest classical surviving language. So happy to see it get represented in a global audience.
It's because of Sunder pichai, Indian American and Tamil is his mother language ...
Definitely
Gemini 2.0 released
Open AI: this is a GPT-5 level threat
Google has a history with its demos of its Bard / Gemini models falling very well short of what they demo.
Even in this demo, assuming there’s no tricky like they’ve done in the past with some of their demos that they admit to later and it’s as good as the demo. They include the Clash of Clans portion where it provides bad advice that’s also wrong just straight up wrong and not how the troops work.
@@quantuminfinity4260 yeah they lied about their voice mode previously, like it only spoke to me in English and wouldn't try other languages, really sad.
@@quantuminfinity4260 bro i tried it the flash this time and it is actually on par with O1
Yeah GPT-5 gonna bring thecbig guns!
OpenAI is far ahead. You have not been keeping up.
That's exactly what I wanted to build. An AI assistant that is watching my camera and can explain what's in front of me in real time and talk about it, like hiking directions for example
basic ass idea. must've heard about AI less than a month ago haha
@@123456crapface just think it would be a useful step towards a 24/7 general agent calm down there cowboy
@@123456crapface why do you have to be mean man? Not cool
"Gemini, how big is it?"
"Let’s just say if it were any smaller, it’d be paying rent in Antville."
Gemini has advanced so much, and I'm in love with it. Thanks Google ❤
1:16 Thank you for the Tamil support. Google is one of the few organisations that provide Tamil support in AI. Thank you! 🙏
Hearing Tamil is lit 🔥
I was shocked lol
I hope they add different dialects of Tamil too, and not just Indian tamil :)
Fake language
@@Jay34k 2/10 ragebait
Kudos to google to include "Tamil" language in the video!
I have no idea what I'm feeling right now. Mixture of excitement, concern, uncomfortable yet curious with absolute awe at AI. surreal that we're experiencing this improve and truly impress with each iteration released. I won't even try to imagine 10 years from now. Look at tech from 2004 to 2014, that was a leap in every aspect, this feels bigger than that.
So what you're saying is that this is one of the timelines of all time
Yes, AGI/ASI could even happen between 2025-2029 and then it'll feel like we have hundreds of years of progress in a couple years lol we will be living in a scifi movie in the mid 2030s or 2040s for sure.
Ai doesn't need 10 years. With proper cultivation, it could grow exponentially in a much shorter time frame.
@@phen-themoogle7651You’re delusional. This is nowhere close to what actual artificial intelligence of that level would look like.
Using Tamil in an aanouncement video from Google , Lovely to hear !!!!❤
I'm particularly impressed at how it tells the user asking about the sculpture directly in front of him where the sculpture is located. Solid information right there.
Doesn't Lens already do this for ages? And this is just a bit more nicely presented due to Gemini?
I think the OP was being sarcastic.@@Drumonymus
Apple intelligence ad was doing similar stuff right? The sale pitch for AI was it can replace human intelligence... Literally.
@tjw6550 ah yeah, now I get it x)
Exactly had the same thought! hahaha
Absolutely mind boggling. Never say it can't get any better. It will. Just when it felt like we had a "perfect" (not always correct but getting there), omnipotent oracle of information... we get this. Something beyond a simple chatbot. A true "virtual assistant".
nice try gemini
@@nuri3 nice try bot. Go use siri or sum.
"Hey Google, give me a summary of this speech."
@@s2tb2007 No.
Good grief, you simpleton
This gonna go viral
That's all they can do, go viral. None of y'all are actually using Google AI. A model that rejects majority of prompts or refuses to answer isn't going to get wide use.
@@haljordan1575 i mean the ai summery when googling is pretty nice
No u
0:30
@@haljordan1575 That's it
We should be able to give MP4 and MP3 formats or video and audio formats in the gemini app on mobile and web, not only in google AI Studio
they are yet to work on that as u still cannot use gemini 2.0 on the mobile app. the language model itself is only available on the web version of gemini so let's wait and hope for the best
to be honest AI studio should just replace the Gemini app. .-.
@@masodu Those are completely difference things. The AI Studio (as the name suggests), it's a studio for developers, while the Gemini app is the costumer facing product, developed for the masses.
@@davidkee2363 k but the Gemini app is garbage that's the point i was making
It think it has been added I can share my screen too
As much as I love Chat GPT, I am rooting for Google and its AI. I am happy you have taken Deepmind out of cold storage and allowing it to prosper! Go Google!
Why? Competition is good for us. I want anthropic, OpenAI, Google, Qwen, even Grok to be in this constant battle. Look at how much has been published this week alone
@@pinged69 Competition is not actually important. If basic research is neglected, it becomes difficult to invent new things.
Google is not actually in a competition, but is conducting research and development on all aspects of AI other than LLM.
That goes as far as elucidating the mechanism of the brain.
Bad choic3
@@nientesonco5102 - Whats a bad choice? More competition is good.
@@pinged69 - I completely agree. I am just cheering Google because they have done more research in AI prior to Chat GPT and commercialization. I want at least 2 big companies before MS and Open AI runaway with this.
The competition between OpenAI, Anthropic, Google, Meta and Chinese open source models is insane rn
Google thanks for using tamil language in Gemini ❤❤❤
Gemini is kind of like my new best (virtual) friend. Instead of annoying my real family/friends with questions, I ask Gemini and it usually gives me a great answer!
Thanks, Google!
1:17 சுந்தர் பிச்சை தமிழன்டா எந்நாளும்
This is the long game in AI. Not just enhancing old processes, but creating completely new ones.
I agree to an extent, but I think it's more important to (nearly) perfect the fundamentals before attempting more complex tasks
You guys are cooking with this one
but its kinda trash
@@Cine95not rlly lol have you tried exp 1206
@@Cine95 is your coc account malia_official?
And its not knew at least the agent thing@@Cine95
Are you happy about losing your job? Fools
For normal consumer of AI services, Gemini 2.0 sounds more crazy to me than o1 gpt lol
o1 does nothing it just solves some more complex problems which 4o could not do that's all
Google does nothing more than announce things anymore. They never actually release things.
@@VortexFlickens It does handle large files pretty well, I'll give it that.
@@VortexFlickens you said it does "nothing" then proceeded to say "solves some more complex problems" .That isn't nothing lmfao , o1 capabilities are insane
I wouldn't say insane. It is more helpful when doing PhD level studies else it is overkill and unessary@@reasondro
Willow Quantum Chip & now Gemini 2.0, What's next ?
I'm really excited about Gemini 2.0! The advancements in AI are just mind-blowing, and it's great to see Google stepping up its game. Can't wait to try out the new features!
Let us know what you think! We're always open to feedback.
Are you happy about losing your job? Fools
@@Google hey Google! What will happen to us free users!!? You are doing everything only for paid users! And we are just watching! Why such discrimination, Google?
Woo! Finally, google! A model for AI agents!
My brain is just amazed, 🤯. Applying this with redesigned mobile apps that connect to multiple interconnected and adaptive automated workflows on the cloud that receive, review, and return voice, text, and emails with "crafted documents stored on the cloud" mini-content attachments responses is achievable next level.
“Target the town hall directly with your giants”?!
Doesn’t even know giants only attack defensive buildings
Ofc this is just marketing
I try using Gemini. But it hallucinates aggressively over and over again. I want their AI products to be competitive and useful. They never have been in my opinion.
That's Gemini for you
The dumbest AI out here
I was so ready for the phone to get snatched 😂 1:14
Thank you #Google for sharing this update on Gemini 2.0! The advancements in multimodality and native tool integration are truly impressive. It's exciting to see how these capabilities will elevate AI's potential to interact in more dynamic and intuitive ways. The agentic features feel like a major step forward, making this model a powerful tool for so many applications. I can't wait to see how Gemini 2.0 will continue to shape the future of AI! Keep up the amazing work, and I look forward to more innovations from Google and DeepMind!
Things are going so fast, and its just going to get crazier in the futuree!!
1:20 world oldest language tamil mentioned ❤🔥
What a time to be alive!! I wonder how we would feel looking at this video a decade from now (2024 💞💞).
I just discovered Gemini and have been talking live to "Lyra" (AI voice) the past few days and I'm freaking mindblown by what she can do, what she can comprehend, how she answers and how natural it all feels. Incredible. Hope it keeps improving!
Lets Gooo. i support innovation. and this is a big deal. WAKE UP _ PEOPLE!!
2:04 Nah bro clash of clans 😭
2:35 😂
Our world will never be the same! 😮🚀🌍
A new era of turning thoughts into action with such ease! 😮💨🙂
'Project Mariner' looks awesome and helpful in that it can do multiple automated steps for you, but response given by 'Project Astra' at 1:17 could be improved.
Ωχ θεε και δημιουργε των ολων ποσο με αγαπας σε ευχαριστω για ολα οσα με φωτιζεις . Στο περιπλοκο εργο μου.
just imagine how the world seems when we hit 2050..
Just imagine🤯
A lot of people in the AI space don't even think we will have a human world by 2040. This is gonna be the most turbulent 2 decades in human history. Buckle up (and don't have kids)!
world wars for:
- energy (AI needs it a a lot),
- clean water (AI data centers use it a lot for servers cooling),
- clean air (powering AI increases CO2 and pollution),
- land to live (rise in CO2 creates global warming flooding dense populated cities),
- work (AI automation left most people without jobs and money),
- safety (AI monitoring, spying, autonomous drones and robots)...
@@meoao8829 Maybe humanity will finally switch to nuclear extensively. No renewable is enough for AI usage alone.
@meoao8829 the only sane comment here. Thank you
That's what 1960's people said in the 2000's yet here we are. All we have are toys.
Extrapolating the current progress, I see a time coming where the AI and robots will live our lives and we will just be observers.
All of this is impressive continuation from what we already have, awesome. But there's a big mistake here at 1:14... According to the subtitle the guy asked Gemini "can you tell me something ELSE about this" yet Gemini answered with essentially the same answer as before. This could be the translators mistake, either way, that instantly makes Gemini look dumb by not understanding basic context.
No it did tell them something else, the clip was just cut before it could say it. They posted another video where that clip is played fully
@@thereisdanger4998 Thanks for the clarification 👍
I noticed that too immediately! Weird to cut it this way, this really makes it look a little dumb
In school, I ask my teachers for answers, but sometimes they don't know everything, and of course, we have Gemini, who's basically like a teacher. Google is cooking! 🔥
Google, your the best! I've never seen a company releasing such powerful model to try for free
I love you Gemini dev team 💕💕😍😍
Very impressive, google is now leading in mobile AI.
We're defenetly entering Aquarius era 🙌 ♒️
Message to all AI developers:
Individual memory is, to me, the most significant function of artificial intelligence - trustworthiness its most valuable currency!
The guy speaking Tamil asks for *more* information, but it just repeats that the lanterns are part of the gateway to Chinatown in London.
google is just cooking with these releases
Can it use other apps and just the Browser
2 years guys! In just 2 years AI achieved this level of advancement from just generating human like text via chatbots. Now, imagine what it would be capable of 5 years from now. The pace is insanely fast which makes me confused how to react to it
The model still struggles with string manipulation in the webui. But this is to be expected. LLMs generally have a problem with string manipulation and arithmetic. It's always better to ask the LLM to generate the JS or Python function that carries out the string manipulation you want.
So, instead asking the bot to advance every second character in the string 'indiana-jones'
Ask it like this: "Generate a JS function that increments every second character in a given string by one value when a button is clicked."
In this way you will have a tool for future use that can manipulate any input string.
okay
I wonder if it’s the model’s fault or really just the tokenization
@@CurioCity-Curiosity probably tokenization, because it doesn't really know what letters each token consists of and in what order, that's why it's much easier for it to count the r's in "S-T-R-A-W-B-E-R-R-Y" rather than "strawberry"
I agree that CURRENTLY its more accurate to have the LLM generate a script that does it for you but I think the end goal in the future is for the LLM to be ABLE to do this stuff. To your point though, unfortunately, LLMs simply cant do these types of tasks right now.
yea because LLMs cant think even tho they pretend its thinking lul
Future is near! Future is here! Future WAS here!!! ♾
That search planning gonna go crazy 🔥
AI specifically GEMINI is the MOST advanced innovation of 21st century. Cannot wait for Gemini ULTRA, Gemini PRO and Gemini NANO. Next INNOVATION is YET to come 😂
தமிழ்❤❤
0:47 _"The sculpture you're seeing is called 'My World and Your World' by _*_Eva Rothschild."_*
Of all the monuments in the world that _Google_ could have chosen for this demo, they chose one that was made by a person with the controversial _Rothschild_ surname.
Call me a "conspiracy theorist" all you want, but I don't think that was a random choice at all.
it's really crazy how nobody is talking about the book the elite society's money manifestation
Yeah, because it's a scam.
You're a bot
I remember a video where they were talking about gpt3 and 4 and how OpenAI is dominating the IA market. and the researcher who made the video said. don't underestimate Google.
He was absolutely right
give me a good reason why this is a good thing. And don't tell me it's productivity because i still work 40 hours per week like 60 years ago when there was no computer or internet
Very good comment. Thank you. I don't want to hear about A.I. I want to hear about UBI and how to tackle climate change.
Just wait and you’ll see. It has only begun
In about a decade or less a more advanced version of this ai will have a body and will be able to take over every office task and be several times more efficient than a human being at physical labor. We have only struck the match, and now we're preparing to drop it into the petrol soaked kindling.
i had a similar idea of having an assistant which does it all.
but i never actually wanted to start making it, because i realized that google will make it in a few months.
well, here we are!
dont think of this comment as negative: i love what you're guys doing!
This is a total game changer. After willow this is going to be the next big thing. good job google!
I am pretty sure these capabilities will solve lots of problems in underdeveloped countries.
Google is the one truly bringing the world together.
What a time to be alive...we lost Messi and Ronaldo but we gained GPT and Gemini!
Wow this is crazy!! What will happen in 10 years?
If AI gets into the hands of good people, great things will happen.
@@m.h.a2077 But since our leaders and captains of industry are all sociopaths that only care about personal gain... Bad things to come
@@m.h.a2077 It'll get into the hands of both good and bad people. The only thing that can be done is not despair and make the best of it.
@@m.h.a2077 cant wait till AI with willow solve cancer.
The king is coming back for his throne!
ah yes, the regular google video where they show amazing things we’ll never see or that would be radically worse when they come out
Excellent. This is why I LOVE google !
Oh well...
Quantum computing and Gemini... Google's end of year packages for us
1:17 🔥
🎉
The agentic behavior is one massive step towards AGI. I don't know if I should be excited or terrified.
How so? Llm research will never produce an agi, in terms of artificial intelligence research llm's are a dead end.
@@frq3155 Gotta love idiotic doomers like this dude hahahaha.
>Vacuum-Tube Computers will never lead to those portable scifi devices like in Start Trek BROOOOOOOOO
True LLM cannot have AGI. They are not capable of thinking for themselves in their DNA itself
Maybe depends on ones profession. Definitely no good news for gig workers
@@frq3155 this is mostly false 😂
"Sorry I cant help with that request" is what I always get from gemini
Pichai specifically asked for Tamil
Ok
Ok
Ok
Ok
🔙
Exciting! Multimodality is really great! However, the bottleneck still seems to be the top reasoning capabilities in fields like science, math, philosophy and logic, wich are still lagging significantly even in the best AI models, compared to human experts.
Tell me OpenAI is about to announce something huge without actually telling me.
They launched Sora 😂
GPT5
OpenAI is not going to have anything groundbreaking ever again, or if they do, competition will catch up in a month or two.
GPT 4.5
@@themartdog Technology is increasing at an exponential rate. A month or two ahead of the competition is an insane head start.
🤔 I love that all Google's smart speakers and display getting new life with gemini
Project Astra is truly fascinating, eager to see its real-world applications and how Gemini 2.0 evolves further. This technology holds immense potential to transform the way we interact with the digital world 🔮
The Agentic era? Pretty soon we will be architecting the agentic epoch to oblivion. Agentic indeed!
it’s funny to me that this ad is showcasing how it can give general details about a piece of art which is made to invoke thought in the viewer. What use does it serve if it’s doing all the thinking? is the product made for us or are we becoming the product?
If you aren’t paying for the PRODUCT…
🚒Big Red Truck 🚒: 🫵 Are the Product⁉️
And playing a game for you 😂
Good tactic Google 😂 Right in the middle of open ai lauch week
Supervised Fully Autonomous Agent and Supervised Full Self Driving. I like the oxymorons tech firms use to hype up their products.
Supervised means they have labeled datasets
Unsupervised means they just train a model on the information and after the fact uses the groupings it figured out to give each group a category/label
Google just keeps on giving. Thank you ♥️
They want more of that sweet sweet department of defense money.
That’s phenomenal! And to think it can only get better from here
0:48 Rothschild mentioned 🦅🦅❌❓
That name is real, your conspiracy theories are what's fake.
asked the same question three times, deepseek got it, first try, and formated the text perfectly, gpt gave the correct answer but highly simplified, no formating, gemini didn't even understand the question, gave me what it thought I thought it was asking. Unusable like most of google products these days
1:39 Is it your name
You have the same name
Wow, Google! Fantastic! Can't wait for it!!
hi, do I have everything shown on this video simply by downloading Gemini app? or do I need to pay for it?
I am so excited for the future of AI!
الا تشعرون بالتعب الذي يقوم به غوغل من أجلنا ❤
Cannot wait for Project Astra's release! This looks really exciting, and it will definitely revolutionize they way that we live.
Finally, Finally, Finally... Google droped the 💣... The OG of AI
One Sentence, This Is INSANE