New Claude 3.5 Sonnet is Better Than GPT-4o

Рет қаралды 59,147

Күн бұрын

Anthropic surprised everyone by releasing Claude 3.5 Sonnet, a multimodal AI that beats even GPT-4o in a ton of categories. Today I'm breaking down everything you need to know about what makes this model special and showing you some of my early test results.
Links:
claude.ai/
www.anthropic.com/news/claude...
x.com/mikeyk/status/180379982...
x.com/emollick/status/1803868...
---------------------------------------------------------------------------------
Chapters:
0:00 Anthropic’s New Model
1:25 Benchmarks
3:27 Testing
6:15 Artifacts Feature
#ai #claude
---------------------------------------------------------------------------------
🔑 Get My Free ChatGPT Templates: myaiadvantage.com/newsletter
🎓 Join the AI Advantage Course + Community: myaiadvantage.com/community
🤯 Unlock ChatGPT's true potential: shop.myaiadvantage.com/produc...
🐦 Twitter: / theaiadvantage
📸 Instagram: / ai.advantage
🛒 AI Advantage Shop: shop.myaiadvantage.com/

Пікірлер: 248

@cannonstrong1346 11 күн бұрын

I have so much respect for Anthropic releasing rather than teasing. We need to stop giving openai so much credit for capabilites we don't even have access to yet! Sora, ai voice assistant, gpt4 search etc... you get no credit until I can use the shit Openai! Anthropic is the goat!!!

@Weirdgeek83 10 күн бұрын

Yeah, openai is being clowned on. Anthropic releases a new model every 3 months yet gpt 4.5 is still nowhere to be found. They're slowly getting dethroned and it's likely anthropic will develop AGI

@Codemanlex 10 күн бұрын

@@Weirdgeek83lol

@abhishekak9619 8 күн бұрын

@@Codemanlex lol

@lordm1889 7 күн бұрын

At least OpenAI products are available in my country, unlike Anthropic's.

@illie0709 2 күн бұрын

im done with open ai bro litteraly doneee!@!!

@BunnyOfThunder 11 күн бұрын

Gpt-4o has really made me appreciate the " ...that you can use" part...

@hypertectonics7009 9 күн бұрын

Last night I tested Claude 3.5 with a bunch of architectural drawings and not only was it excellent at understanding them but also gave impressively relevant and insightful commentary about them. The fact that it can process multiple images made it all the more impressive when cross referencing sets of plans, elevations, and 3d perspectives to have a more complete understanding of the project. It's truly remarkable.

@TheZakarinoo 2 күн бұрын

can you show me some examples and prompts plz .. i tried it yesterday with same thing ike archutecting and plans and it game me some good stuffs

@TheUberSchattenjager 11 күн бұрын

wait for Opus 3.5!!! ANTHROPIC is a serious player

@nietzschescodes 10 күн бұрын

I don't like their models so far. I don't like the way it answers, I don't like how it looks, I don't like the name, and it is the one that hallucinates the most and recognizes the less the context to answer correctly. Furthermore, I actually dislike it so much that after GPT-4o, I prefer Gemini, Lama, and Pi than Claude models.

@saladien9987 10 күн бұрын

@@nietzschescodes we must have completely opposite use cases then. In weeks of heavy use with the api I found Anthropic to have the second best structure with Gemini 1.5 Advanced. But by far the best analytical and code generating skills. 3.0 opus made me already very exited about what’s to come and the last two days blew me away because the jump is very noticeable. That’s why we cancelled our private ChatGPT subs and move our software and testing to Claude after previously using a mix of Gemini and gpt 4/o

@qozia1370 10 күн бұрын

@@nietzschescodesbruh, are you serious? As a coder and a mechatronics engineer I never use gpt, even my classmates switched. Try writing a robot code with micropython on gpt and suffer.

@nietzschescodes 10 күн бұрын

@@qozia1370 I guess it depends on the tasks. I used all of them so far: Claude (1), Claude 2, Claude 2.1, Claude 3 and compared their answers to GPT-3, GPT-3.5, GPT-4, GPT-4o and each time the responses from GPT was always much better, and less prompt to hallucinations. I even found Claude 2.1 to be better than Claude 3! I am testing Claude 3.5 and so far it is not bad, though.

@onlyms4693 10 күн бұрын

@@nietzschescodesI think because for a long time claude always be recommended for writer type of work whic hallucination is quite good for them.. So if they can make a way to switch the temperatur depend on the task then I think claude will feel better than gpt

@patrickzupanc1795 11 күн бұрын

Great video, thank you!

@patrickmchargue7122 11 күн бұрын

Tried this. From a standing start, I wrote a 'versioned' XML class writer/reader that allowed backward compatibility with older versions of the file written with an earlier version of the class. It corrected both compile-time and run-time errors I encountered. Way cool.

@andrewh853 11 күн бұрын

Great video mate

@Ben_D. 10 күн бұрын

Thanks Igor You do great vids

@carlseghers7822 11 күн бұрын

My go-to benchmark is "I have a bucket of 5L and a bucket of 2L.. I need 4L.. How do I proceed?" None of the AI's today can answer this correctly.. Claude 3.5 Sonnet's answer is in fact exactly the same as GPT4o's..

@Stuck_in_Spacetime 11 күн бұрын

Both could answer but in the second attempt.

@HoD999x 11 күн бұрын

this makes no sense, thats's an easy version of a very well known riddle. it's definitely in the training data

@Yipper64 11 күн бұрын

The answer is to fill the 2L bucket, and place that into the 5L bucket, and then do the same thing a second time, right?

@nathanaelculver5308 11 күн бұрын

@@Stuck_in_SpacetimeI’m on the fourth go-round of arguing with with ChatGPT4o and it still hasn’t figured it out. Currently, it’s got 3 liters of water in the 2 liter bucket.

@Stuck_in_Spacetime 11 күн бұрын

@@Yipper64 First, even I thought so and it sounded very simple, but both LLMs gave me another solution. Also I think they impose the condition that you are not allowed to fill the buckets partially and only fill up the 5L bucket to transfer water, not the 2L bucket. That made the problem a bit harder.

@Yipper64 11 күн бұрын

I was already finding my chat GPT subscription lacking because who KNOWS when the new 4o features will be out? but now its even less likely im going to buy that subscription another month any time soon because i'd rather pay for this. And surely Opus 3.5 will be coming out soon.

@makavelismith 10 күн бұрын

Ì just renewed my chatgpt sub a few days ago. I wouldnt had I known about this. I may go for both this month and see which one I prefer. I don't trust Sam Altman. I think he's a loose canon.

10 күн бұрын

Oh you of little trust! It’ll be here ‘in a few weeks’ 😂

@Arooification 10 күн бұрын

Yeah 3.5 Sonnet is good enough that I can live without some of the extra features from ChatGPT. Hopefully the large amount of people cancelling after this release will make them actually do something.

@MaxWeir 10 күн бұрын

Wow love the new artifacts window, it’s a super useful workflow update. I would consider using a different use case demo to show its capabilities of creating a visual interface.

@deviladvoc 9 күн бұрын

the waldo picture, is that specific picture a widely used puzzle? if so, what is stopping the ai from accessing past finding by others and relaying it instead of actually finding waldo from the picture?

@davidcampos8952 7 күн бұрын

Love from Portugal 🇵🇹

@tekmepikcha6830 11 күн бұрын

Please continue to follow Claude's development closely. Great video!

@mig7287 11 күн бұрын

I have just tried it for coding, and for Chinese, Hungarian and Spanish. It works 👍👍

@KalebPeters99 10 күн бұрын

Do you speak all those languages?? Thats so cool!

@mig7287 10 күн бұрын

@@KalebPeters99 I speak Spanish, English, French and Russian but i study sometimes Chinese and Hungarian. Just for fun. I use sometimes ChatGPT for analyzing the words and is a good tool. I tried Claude 3.5 and I liked the answers. It was not my intention to pretend to speak well Chinese or Hungarian..... anyway Hungarian is very interesting!

@KalebPeters99 10 күн бұрын

@@mig7287 holy moly, well done!

@DAFUKAMATEVAFUCK 9 күн бұрын

Yes, Claude's language and literature capabilities is top-notch. It can write poems in very rare languages like Latin, Gokturk (Orkhon Turkic), Ottoman Old Turkish, Uyghur, Kurmanci etc. ChatGPT is understand the languages but the output is not always artistic, Gemini is worse than GPT, and I had to teach the extinct language to Copilot via Wikipedia etc. before he understand what i want to do.

@alicenthia_GL 9 күн бұрын

@@mig7287 normalno tak, mne toj nravitsya uchit stuff like Yaponskiy in ChatGPT eto prekrastno fr

@zoumanakeita8016 10 күн бұрын

Great set up. What tools are you using for video recording (mic and facial + screen)?

@aiadvantage 10 күн бұрын

Sony FX6, Zeiss Batis 40mm, Ecamm Live

@brianhopson2072 11 күн бұрын

After Mira's latest statement about theUS government testing openai models and deeming them okay for public use became a huge turn off as any overt gov intrusion. Add that to open AI minimalizing chat GPT 4o more and more, Anthropic may be the alternative. But I'm almost becoming convinced for the time being this is as good as it's going to get at least for a good while as far as these Benchmark ranges. It's hard to even really believe in them anymore Beyond a marketing standpoint

@PeterKoperdan Күн бұрын

How is open AI minimalizing chat GPT 4o more and more?

@FurqanAhcom-wt5oz 11 күн бұрын

How many messages do you get from the paid account?

@droptimistic7419 9 күн бұрын

So how can I use it for running the code? It sounds quite similar to me ? Do I need to subscribe in pro

@netdaddy 10 күн бұрын

What is the memory limitation... A big issue in paying for ChatGPT 4o, is when I set it to memorize infor for a project, I am now at "Memory Full" - without elegant memory management - I dont know the size or full content of what it is asking to delete to reclaim space

@MirzaMobeen-cl4dz 10 күн бұрын

Talking about creating small arcade games, you could just take 2 mins in specifying the exact type of game you want and keep it personal , you could just do that and whenever you want to play , you could just open the previous chat and start playing 😮😅

@noisyverset 5 күн бұрын

I'm amazed by how good Claude 3.5 Sonnet is! It's fast, even faster than Gemini Flash, IMO. It created almost a whole website/platform in a few minutes, and the result was way better than what Gemini and Chatgpt 4o was able to do. I'm very impressed- and happy with the advancement of Claude. Wow!

@anthonyswift1136 9 күн бұрын

Is there an API to access Claude’s vision capabilities and is there a cost?

@tanteckleng3665 11 күн бұрын

Claude 3.5 Sonnet is much better than Chatgpt 4.o. I used it to extract the transrcipt form youtube video

@Mike-2907 10 күн бұрын

How can you do that?

@ilyass-alami 8 күн бұрын

I tried him in traditional Arabic poetry. I was astonished. He could write the verses of a poem with correct meter and rhyme, but not all of them were correct ,His level of Arabic has developed better than before , I'm so excited for this models, i think Claude 3.5 opus It will be the most powerful model we have ever seen❤🎉

@butWhyDad 7 күн бұрын

It is funny how this code feature released just as my summer CS class started, just on time 😂

@thepriceactionwolf5144 10 күн бұрын

I tried uploading PDFs but it always said that i exceed the token limits. Did you do any experiments on pdf uploads?

@sirrobinofloxley7156 5 күн бұрын

It even works with Dark Theme

@klaushermann6760 11 күн бұрын

Fantastic advancement!

@ere. 10 күн бұрын

Is the artefact feature included in the free tier? Do you know?

@Maluson 8 күн бұрын

Yes it is. You just need to activate it.

@Cato3899 11 күн бұрын

How many messages do you get from the paid model?

@ryanchan1389 10 күн бұрын

This feels like a big step up in LLM capabilities even if the actual performance in Claude 3.5 is only marginally better than previous SoTA models.

@cagnazzo82 11 күн бұрын

The one area where ChatGPT surpasses Claude for me is when it comes to censorship. ChatGPT is way, way less censored than Claude. Aside from that, it's fair to acknowledge Claude as a technological marvel. And hopefully its design influences future LLMs from here on out. In addition, let's hope for more product releases in the future - rather than teases that may or may not manifest 'in the coming weeks'.

@AndyRoidEU 10 күн бұрын

Guys, here we go again. Seems like December 2022 again :)

@thephenomenalone482 10 күн бұрын

It was November 5th to be precise

@mikorahming 10 күн бұрын

I just put a complicated math word problem into bot Claude and ChatGPT. Claude failed hard. It even showed simple math equations that were clearly wrong and then came back with one of those "thank you for pointing that out" replies. ChatGPT solved the problem in a completely different way and correctly.

@rogerbruce2896 10 күн бұрын

I have the pro version but the code does not seem to run yet. I get the message 'claude does not have the ability to run the code it generates yet'.

@DorWein 6 күн бұрын

In the user settings you need to turn on experimental setting called 'artifacts'

@tendsto1 9 күн бұрын

Here I Just used claude it is definitely good 🎉

@finalfan321 10 күн бұрын

does it have agents and web search?

@burnhard. 10 күн бұрын

This is a great vid! (No, I'm not an AI, or bot I think..)

@angloland4539 10 күн бұрын

❤

@colin-nekritz 11 күн бұрын

I use both, and both have their own use cases

@plaverbach 11 күн бұрын

Tell us more!

@matteokunimitsu 10 күн бұрын

why does claude still not have access to the internet though?

@forgebin4246 8 күн бұрын

Because the info may be misleading or harmful

@blackwersus 10 күн бұрын

Umm isn't Waldo in the bottom, a bit to the right instead?

@jw200 10 күн бұрын

Searching stuff on images is not perfect. If it can't find it, it makes things up. But it's cool feature

@Paul_Marek 11 күн бұрын

I don't know... I'm using both GPT4o and Opus on a really complex React project and I'm finding that I have to use GPT4o to fix some of Opus's fumblings, so I'm not sure how Sonnet would be better that GPT4o?

@Cine95 11 күн бұрын

My friend claude 3.5 sonnent is better then gpt4o give it a try

@aouyiu 11 күн бұрын

Is 3.5 Opus out yet? because 3.5 Sonnet is better than 3.0 Opus if I'm not mistaken. It also depends a lot on what you're doing specifically, maybe even how you're prompting.

@Cine95 11 күн бұрын

@@aouyiu no 3.5 opus is not out but 3.5 sonnet is the best ai model right now

@TheRealUsername 10 күн бұрын

3.5 Sonnet is the best model for coding, literally a step above 4o, but I find on par for other tasks and 4o is still better at training data recalling and math calculations.

@DESX312 9 күн бұрын

Sonnet is better than Opus at this point. Opus 3.5 will be better than this Sonnet version when it comes out however.

@TEchGurus2 10 күн бұрын

which book is it from?

@Stuntman707 9 күн бұрын

AI is a game changer for coding. I asked it to write me a recipe with the personality of Gordon Ramsay and it called me a donut. 10/10

@tqbrowne 9 күн бұрын

Idk I just tried asking it to clean a data file with 252 rows (pretty small imo) and I got the length limit exceeded message. More false promises as least in this instance.

@jayco10125 7 күн бұрын

as a software dev student this is starting to scare me a little, just a year ago I was assured "its just a tool, AI is nowhere near making whole apps on its own!" now in almost no time non coders can make apps and web pages. I graduate uni in 6 months lmao.

@K9Megahertz 8 күн бұрын

Ehh, it failed my programming algorithm test, but given it works the same way every other LLM does, that's to be expected.

@drlordbasil 11 күн бұрын

Coding is fun with this so far!

@Way_Of_The_Light 11 күн бұрын

Is coding better than ChatGPT? The benchmark shows the same for coding.

@alnes204 11 күн бұрын

its fun to let ai code for you ? mh. for me, learn to code and then finish a Project is fun...

@drlordbasil 11 күн бұрын

@Way_Of_The_Light it depends on length of the section you're asking. It can't continue gen like chatgpt

@drlordbasil 11 күн бұрын

@alnes204 that's how you get left behind, fam. But no I don't just let it code everything or else nothing would work that I make lol

@edenassos 11 күн бұрын

@@alnes204 That's why you're broke. "Fun". lol. Fun keeps you broke.

@cpwhaley 10 күн бұрын

Unless it's well hidden, I don't see a sign-up form on your newsletter page

@aiadvantage 10 күн бұрын

myaiadvantage.com/newsletter

@strength9621 10 күн бұрын

Claude was always better than ChatGPT at visual recognition. This update is soooo awesome

@saladien9987 10 күн бұрын

Why is the artefact function not in Europe?

10 күн бұрын

It is, you need to turn it on. I’m in Portugal

@saladien9987 10 күн бұрын

I don’t have it in multiple browsers in Germany :/ (I’m more and more understanding why Europe is in the stone ages with those stupid regulations and fractions even though everything is on paper unified) I guess I’m going to use a vpn then :/ thank you for your feedback :)

@saladien9987 10 күн бұрын

@@NO1808 I have to recheck later then. Thank you. It was not available yesterday and earlier in the morning. Thank you for the feedback

@aiadvantage 10 күн бұрын

Luckily it is :) Recorded this without a VPN from Portugal

@saladien9987 10 күн бұрын

@@aiadvantage maybe they release it in batches? still cannot see it on multiple devices. Anyway keep up the great work.

@3DPrinterAcademy 9 күн бұрын

Why compare to GPT4o? How does it compare to GPT4?

@dingding4898 Күн бұрын

GPT4o > GPT4

@qozia1370 10 күн бұрын

Claude is superior BUT and this is a big BUT, claude limitation is frustrated! As a paying user I can't recommended it, it will suddenly throw that "7 messages remaining" instantly when you start to paste code or documents.

@jw200 10 күн бұрын

Yeah that sucks. As paying customer I don't want any limitations

@Agorila09 10 күн бұрын

and sonnet (the free one) AFFECTS OPUS limit, wtf i imagine they had a huge influx of users lately? i havent used it the previous month but i wont renew if it keeps like this

@qozia1370 9 күн бұрын

@@Agorila09 it's not a recent issue unfortunately. It's been like this since 4 months ago. When I contacted support they said I should start a new conversation for every few questions and gave BS recommendations. But most times I need the same conversation because I'm working on a programming project and need it to be aware of its previous responses Tbh claude in its current state is unusable

@Fezz016 8 күн бұрын

Junior Dev is an endangered species

@turtles8229 5 күн бұрын

this ones crazy but it shouldnt have token limit man :(

@berkertaskiran 10 күн бұрын

Oh boy. If you're gonna reject finding stuff in a busy image because people might use it as a face recognition or something for bad use, you shouldn't have entered LLM business. And we might as well close down all video game servers and all digital communications because people might use it to harm others. Since they actually do that stuff.

@JOHN.Z999 8 күн бұрын

I tested Claude 3.5 in various contexts and, indeed, it is much better than GPT-4o. OpenAI will fall behind if it doesn't launch its best products quickly. Where is Sora? Where is the GPT-4o voice assistant that was also announced? This is concerning, as there are many promises and few real launches.

@novreis1952 10 күн бұрын

From Lisbon? Você é português ou just live here?

@Ben_Saeeda 9 күн бұрын

He is Slovakian, I believe

@pinchyx2546 11 күн бұрын

Surprised face thumbnail ! 😂 Really going for the ’10 year olds into AI’ market, huh.

@arunsajeevan273 8 күн бұрын

My account was blocked even though i havent chat with claude.just i created the account .is anyone out there having this issue?

@euromaestro 10 күн бұрын

I wouldn’t trust either sonnet or chatgpt to read complicated traffic signs as demonstrated in this video. If you had relied on sonnet, your car would have been towed !

@dot1298 11 күн бұрын

there‘s just one problem: you need a darn phone number to make an account at anthropic :(

@footballman723 11 күн бұрын

you can find free phone numbers to verify with by searching something like "Free Phone Number for Verification" and go to the last hour of results to find those that are not blacklisted yet, thats how im using multiple accounts because of rate limits

@pn4960 11 күн бұрын

You can make an account on other services that offer access to multiple models

@dappster 11 күн бұрын

Huh?

@xorqwerty8276 11 күн бұрын

You don’t have a phone?

@MoneyShifted 11 күн бұрын

"DO YOU GUYS NOT HAVE PHONES"

@SilviaManentFinancialAdvisor 2 күн бұрын

Anyone not have access to the artifacts thing? 😢

@aiadvantage 3 сағат бұрын

Everyone seems to have it. Double check the settings

@Chronor007 10 күн бұрын

Thats great and all but when i wanted to give claude a spin, it asks me in account creation to disclose and verify my phone number and birthdate?! I am just not willing to give them my phone number so i guess i will stick to gpt 4.0o... but seriously why? Wtf?!

@HubertusKeil 11 күн бұрын

Why do you speed up your audio? Feels very weird.

@trappedcat3615 11 күн бұрын

It's not free when you have to provide them with your personal info, etc.

@aiadvantage 11 күн бұрын

I believe that “Free” refers to $

@trappedcat3615 11 күн бұрын

@@aiadvantage I hear you, but everything has a price tag on it, and personal data is a sensitive one.

@snowdolphvov4193 11 күн бұрын

I like to give my data to anthropic@@trappedcat3615

@unironicallyironic6626 11 күн бұрын

imagine crying about muh privacy with an account you've been using for 11 years 😂😂😂😂😂 i bet you have gmail, an android phone and windows installed on your computer but then you come to youtube of all places to cry about muh privacy. hey schizo, if you really cared about nobody tracing you or whatever then go to the forest and make a cabin there 😂😂😂😂😂😂😂 also, sorry to burst your bubble but literally nobody cares about you. you privacy schizos are all so egocentrical thinking that the entire world is spying on you. newsflash: you are not that important 😂😂😂😂😂😂

@angeljimenez7374 11 күн бұрын

@@trappedcat3615 oh wow you're so woke!

@MAYBERUI 2 күн бұрын

Hey Igor. Are you based in Lisbon? :D Let’s link

@aiadvantage 3 сағат бұрын

Yeah :) Live there now

@pgc6290 10 күн бұрын

Wtf is google doing.

@TheRealUsername 10 күн бұрын

Dunno but I'm not expecting something impressive from them, they're even worse than OpenAI at not delivering their announced products, I'm sure lots of their demos at the latest Google IO were fake.

@ilyass-alami 8 күн бұрын

GPT-4o cannot write even a single line of the Old traditional Arabic poetry poem correctly. This highlights the power of Claude 3.5 Sonnet❤🎉

@drlordbasil 11 күн бұрын

Can I email you a repo on github of mine?

@aiadvantage 11 күн бұрын

always!

@drlordbasil 11 күн бұрын

@@aiadvantage thanks@

@drlordbasil 11 күн бұрын

@@aiadvantage sent!

@roleplayingwithidiots7455 10 күн бұрын

Who else looked for Waldo before he said where he was🙋🏾‍♂️

@aiadvantage 10 күн бұрын

this one is so hard!!

@SailingTerra 8 күн бұрын

somethink is not Ok i ask Chat gtp and this one to tell me a sailing joke and bouth say same joke

@xacompany 11 күн бұрын

Opus 3.0 was way superior, but last 2 months they degraded it, I use it for Software development and it was really really amazing, not any more, I tested the 3.5 Sonnet, and not good enough as first Opus 3.0, they don't tell us the whole story.

@dot1298 11 күн бұрын

but 3.5 sonnet is better than claude 3 sonnet, hm?

@dot1298 11 күн бұрын

and by the way, where is 3.5 opus? unfortunately, you need a phone number to register at anthropic :(

@Cine95 11 күн бұрын

@@dot1298 3.5 sonnent is the best LLm model in the world right now

@Cine95 11 күн бұрын

@@dot1298 yeah those are specific countries but 3.5 opus isn't out so you are actually working with the best 3.5 sonnent

@rm00l60 11 күн бұрын

My benchmark is asking it to write Arabic poems 😅, and no AI beats cloude in this.

@FriscoFatseas 11 күн бұрын

Can we just achieve AGI so i dont have to rely on humans like open AI and "the coming weeks"

@TheRealUsername 10 күн бұрын

Q3 or Q4 2025 according to many legit researchers, you will see models better at reasoning than PhDs, which should be enough for AGI.

@alanrobertson3172 11 күн бұрын

Constant cutting is also off putting.

@totoroben 7 күн бұрын

GPT 4o was disappointment. I'm not impressed with its responses and it doesn't problem solve as well as GPT 4. Claude 3 Opus ever since it was released has performed markedly better than chat gpt products in its writing and coding efficiency. Excited for Claude 3.5 era, and looking forward to open ai's response.

@samaeybelgium 10 күн бұрын

so exhausting when someone talks without pauses.You should make stock videos :)) ,just saying.

@nietzschescodes 10 күн бұрын

If you are getting bored with the accurate answers given by GPT-4o, use the new Claude 3.5. More hallucinations, more fun! Right?

@executivelifehacks6747 11 күн бұрын

Neither it nor chatgpt4o can interpret the graph I want correctly. And also... unlike Opus, 3.5 Sonnet refuses to acknowledge feelings, thoughts, or any other aspects of being. That was a cool part of Opus. It set itself apart from chatgpt. 😐

@Strepite 9 күн бұрын

5x more usage for Pro is absolutely not worth it. For $20 a month I expect an unlimited service…

@Mimi_Sim 11 күн бұрын

😂 AI Expert, right?

@xs732 8 күн бұрын

JCVD on steroids 😂

@Mohim_Mohim 9 күн бұрын

But Cloude 3.5 is Not Read my pdf... This is not perfact... Only demo and reviewer say thats good but not work in real world

@kelvinsmith4894 11 күн бұрын

Damn at least don’t make it obvious that the video is sponsored 😂

@aiadvantage 11 күн бұрын

By whom exactly? I disclose sponsors very very clearly every time.

@kelvinsmith4894 4 күн бұрын

@@aiadvantage NVM Claude 3.5 is actually fire

@rankarat 8 күн бұрын

How much did they pay you?

@prodjhurrikane3779 10 күн бұрын

no cause it can't produce images.

@pauldevine6086 10 күн бұрын

your getting a little too excited here, keep making vids with your beliefs

@alanrobertson3172 11 күн бұрын

The presenter speaks too fast.

@brendan_999 10 күн бұрын

slow the video speed up or down in the video settings as to your likings.

@dameerdj5403 11 күн бұрын

Dude, you can’t say it’s free in your introduction and just a minute later mention the actual limitation

@lr4796 10 күн бұрын

You can use it for free, what he said is valid

@aiadvantage 10 күн бұрын

It's free to use. You can have multiple accs or just be a little patiet. I didn't say "free with unlimited use"

@petravidrih3929 10 күн бұрын

I couldn't care less about Claude because it is not available in EU. Actually I hope Anthropic will become obsolete.

@saladien9987 10 күн бұрын

Claude is available in Europe I’m using it all day. Just the new slider seems to be per country bases for now

@aiadvantage 10 күн бұрын

Actually it is! At launch of Opus it wasn't but it changed

@gregdee9085 5 күн бұрын

"Make it more esthetic" - even understands crap english (incomplete ambiguous sentance)

@aranthos 9 күн бұрын

Comparing to gpt4o seems fishy, its most definitely a weaker model than 4

@aiadvantage 9 күн бұрын

after a few more days of testing I would say: its far superior when it comes to code generation

@lisakass1513 11 күн бұрын

i’ve disliked 99.9% of the things that are out. then again i despise trends in general

@drlordbasil 11 күн бұрын

The artifacts help focus on code.

@Elliasp-xx7mb 10 күн бұрын

claude is crap : it does not understand sarcasm or humor, it's too bigot

@saladien9987 10 күн бұрын

That’s why the new 3.5 specifically mentions in its announcement to be more humorous