Claude Just Won The AI Wars...
Вставка
- Опубліковано 27 вер 2024
- Claude Sonnet 3.5 IS INSANE - It's more powerful than Opus, More intelligent than Opus, faster than Opus, and.... FREE? This is completely crazy and is an amazing move from Anthropic to beat out its biggest competitor, chatGPT.
Link to harbor:
incomestreamsu...
Sonnet 3.5 is a significant improvement to Claude Opus, which was already the most intelligent model on the market, making this by far the best AI Language Model (LLM) to use -
This model is INSANE and we needed to share this with you ASAP
Thanks for watching... PEACE
Hamish
Prompt:
EMBED IMAGES, EMBED INTERNAL LINKS WITH MARKDOWN Write a 1000 word article. Write 8 titles and 1 paragraphs per title, with formatting like tables, embeds, internal links and lists. Do not use any placeholder information. Make the content about the company - do not invent links, use at least 10 embeds and 10 collection internal links per article. Please generate an article about the keyword "Kiton: The best brand in the world?" based on the following information: Please be creative when writing. Focus on writing good e-magazine style selling content but from the perspective of the business. I have these products - I'm writing for 2men.it. These are all products they sell, and collections of products they sell. I want you, by embedding images, to make a rankable SEO-Optimized listicle article about the keyword. You should use a lot of tables and other formatting. Do not use complicated language. Don't use crazy language, but give a lot of interesting formatting and make it like a journal style advertising the specific products. In the article, please embed some of these collections. I have included the collections in the prompt as well. Please use headers such as # h1 on the top header and then ## h2 and h3 ### throughout the article and be considerate about where you place both the collection embeds and the collection links, and that you use appropriate SEO-Optimized anchor text for internal links.
Join this channel to get access to perks:
/ @incomestreamsurfers
I would be hesitant to think anyone has "won the ai wars" yet. Seems like every day someone else is taking the lead, and it's just the beginning.
One could say... won, for the time being.
@@byronhorde5892yeah I'm pretty sure he's pretty pretty sure of that, more so than the rest of us are.
@@Kelvinapplegatethe surety of the sureness
It's a standard click bait title
Thanks for covering this!!
Thanks for the update, I wasn't aware that Claude was so close to GPT-4.
I'm gonna try it out and see how up to date it is with programming knowledge.
Btw, video ends at 16:40 and there are 4 mins of black screen after that.
Thanks just fixed it - could you email me incomestreamsurfer@gmail.com
I love Claude Sonnet, use it all the time. The only thing I find lacking is that it can’t read and analyze images, you have to attach one at a time, otherwise it mixes them up
Bizarre, when I used the same instructions I got:
"I apologize for the confusion, but as an AI language model, I don't have the ability to directly render or display websites. I can only provide code and instructions. To actually render and see the website, you would need to:
Set up a development environment on your local machine
Create a new React project
Implement the code I provided
Run the project locally"
yes yes haha, it doesnt know what it can do. tell him that it can since the demo showed it and it will do it
Turn on artifacts
Awesome video as usual. Thank you.
I may slightly disagree that Claude Opus is generally better than GPT 4 at coding. We build apps with Java and SQL, and Opus often struggles to solve certain problems after even 6 prompts, then gives up and asks for more context, whereas GPT 4 usually gets it right after roughly the second prompt or even first.
Nobody wins “PERIOD” - its an “Infinite game”
I tried the React thing and it specifically states it CANNOT run its own code.
The thing that sucks with Claude (and I love Claude) is the low usage limits. I'm using PRO and I constantly run out of usage tiers. I'm getting so much value out of it, I would pay 2.5x for 2x the usage tier. But the only possible upgrade is from individual pro ($20) to team with minimum 5 users at $30 each ($150). That is too steep. The alternative is the APi which is great, but would lose the 'artifacts' feature in claude 3.5 which is too amazing for workflow.
Just use the playground when you run out, or have two accounts
Thank you so much
You definitely drop the 💎💎💙... so are you switching over to Claude from gpt? I got a few of those seo gpts and spreadsheet tools you made, are they obsolete now?
OpenAI waits for releases like this and then comes with something better lmaooo
But opus was already better than 4o.
@@conrayccu8390 lol no
@@conrayccu8390 well tbh not really. better for humanistics but not math or physics and coding is same tbh (open ai said it's better but tbh it is same). also here is the voice that will come so it is better. but sonnet 3.5 is something else, i can say that
@@conrayccu8390they were quite balanced
Heads Up, I clicked on your link to your Harbor software, and I immediatly got a redirect to a dodgy url. You might have a click jacker script hijacking your site.
Fixed it I think thanks
Does paid Perplexity Pro version give access to this?
yes
Your website has got a dodgy redirect on it.
Thanks thought I fixed but just checked again
Are you actually serious to claim they’ve officially won the AI war?
First since when there is an AI War? And how do you define a winner to a race without finish line? 🏁 or is the finish line defined as the first to reach AGI? Or ASI? Or what ever else will come in mind?
Next week or next month, we might see a new model more performant than Claude 3.5 Sonnet!
Jesus it’s just a title m8
I guess winning the race in terms of who is on top right now, during the recent updates of all the major AIs. Its too early to call it the OpenAI killer, but fair enough to say Anthropic won the current race.
I closed thd video on "claude has won"
I trued, its dubjrctrcly better but ehen i thriw garder things ag it fails the same as gpt4
Artifacts are very cool. But still need to be better. I am staying with chatgpt subscriotion as customgpts are more useful still.
dude.... You don't care about security and privacy????? wtf???
Bro, make a video.
I don't take his remark literally. Too often our fears are self-fulfilling - just for example, see Japanese toilet paper incident from decades ago.
yehhh full production. welcome new guru
I'm not sure you're reading those graphs correctly, as if you are correct should the sum of all the models not equal 100%?
Hey Hamish, I just tried the same thing, with the React Website - but to me the model answers that as an AI assistant, it cannot render code :-) How did you get that running?
It depends on the code, some won't render in the preview. Easier to just setup a small nextjs site on your local system, and copy the code artifacts in.
Hi Hamish, great video thanks. Also thanks for dropping the harbor link, though the link doesn’t seem to be working right now?
I love Sonnet 3.5
It's great but not as function at looking at files and projects you can upload to it.
When you snickered about safety and policy, I then realized we are soulmates 🤣
Fantastic achievement.