New Claude 3.5 Sonnet Beats GPT-4o
HTML-код
- Опубликовано: 19 июн 2024
- Access courses on ChatGPT and 50+ AI tools here: bit.ly/skillleap
Anthropic has released Claude 3.5 Sonnet, a new AI model with enhanced intelligence and twice the speed of its predecessor. It is available for free on Claude.ai and the Claude iOS app, with higher usage limits for paid subscribers and API users.
Key features include advanced reasoning, coding proficiency, and superior visual processing. The new "Artifacts" feature allows real-time content editing and integration.
Key Features and Improvements
Performance: Claude 3.5 Sonnet delivers double the speed of its predecessor, Claude 3 Opus, making it ideal for complex tasks such as customer support and multi-step workflows.
Advanced Capabilities: The model excels in reasoning, knowledge, and coding proficiency, and has improved understanding of nuances and complex instructions.
State-of-the-Art Vision: Claude 3.5 Sonnet leads in visual reasoning tasks, such as interpreting charts and transcribing text from imperfect images.
Artifacts Feature: A new “Artifacts” feature on Claude.ai allows users to generate, edit, and integrate AI-created content like code snippets and documents into their projects in real-time.
You can try it for free at claude.ai/chats
Read the full announcements here: www.anthropic.com/news/claude...
Access Top Courses on ChatGPT and 50+ AI Tools on Our E-Learning Platform: bit.ly/skillleap
Access Top Courses on ChatGPT and 50+ AI Tools on Our E-Learning Platform: bit.ly/skillleap
It keeps getting better and better.
Wow, this is incredible! Claude 3.5 Sonnet looks like a game-changer. The speed, capabilities, and free access are truly amazing. I'm so excited to try it out for myself. Thank you for sharing this thorough overview and comparison. Can't wait to see what you discover in your head-to-head video!
Looking forward to your head to head video.
I am still testing it - results are quite impresive :)
yea pretty good so far.
@@SkillLeapAI and is available in Europe from day one:) nice change
@@micbab-vg2mu oh nice. They rarely do that.
All lies
3.5 Sonnet is much quicker than 3 Opus, but in my side-by-side tests I'm finding 3 Opus is still far more detailed and useful for summaries and brainstorming. 3.5 Sonnet seems to stay brief, missing out on extra details that Opus gives. Same when asking for creative hooks, promos, and social descriptions that still look pretty basic compared to Opus. I've not tried coding and the artefacts stuff, which does look good. So if a 3.5 Opus is forthcoming with those features, sweeeeet!
Claude is so neat, also a big thing is it doesnt look AI written often, I create GDDs using this..
I subscribed to Anthropic 7 months ago and still can't give it up for another service - for any professional is so worth the monthly price.
Definitely 200k limit is the main pain currently - breaking it would be a massive improvement.
Also the interface could be MASSIVELY improved by small features - voice input, search into the conversation, automated elaborated summarization into the new conversation etc.
Why not offer project knowledge base management for everyone?
Etc. etc.
I did the snake game in html, css, Javascript and I played the game inside the snippet. I don't think it's just set up for python testing yet. I even created a 3d site and the 3d model rotated and I could navigate between screens
waiting for that to drop on brazil
Been using it all morning. Only issue is that I had to get 2 accounts for my workflow because the message limits. I use it for programming and it was getting most of my issues done in 1-3 attempts where GPT4o often takes 5-10 or may never get it. The one weird thing I'll say is that it gave me a typo in some css which was really weird...an actual syntax error. But I'm considering dropping OpenAI this month for the first time since ChatGPT launched.
They appear to be less gimmicky and more advanced. Love anthropic
problem i have is it cant save as pdf, word, excel
in gpt4o i can get results converted to a nicely formatted word document
Claude is super restrictive however. I seem to use my limits within one or two short conversations. “You can chat again at 4pm” message appears. 4pm is often hours away.
ChatGPT never seems to hit any limit.
I prefer Claude for quality output but frustrated with limitations. (In paid sub)
Cool good to know. So how many messages is subscribed claude 3.5?
It is the best model, but a bit too censored for my taste.
The main issue with Claude for me was that even with subscription - I keep running out of credits - and it wastes a lot of credits on easy mistakes or repeating the same thing over and over.
Whenever they fix those issues - I will come back to it :D the new UI is sleek and I like the side by side code and the copy button at the bottom. I wish ChatGPT would move their copy button.
Yes I find the same. Drives me mad.
I'm sorry, you need to reword your comment and try again, as it might be inappropriate and against my fee-fees. Is there anything else I can help you with today?
Hi there. If i want to get the best ai for answering emails some content writing and am only willing to pay for one subscription, which service do you suggest
I would suggest ChatGPT if you had to pick one
Perplexity as it gives you access to multiple LLMs
So it's basically code interpreter 😅 but when are we gonna get web search capabilities
The one thing stopping me full time in Claude over ChatGPT is the ability to make pictures
Just use Bing image creator, it’s the same Dalle 3 AI. There are some things that it won’t make that ChatGPT will, but there are also a lot of things chat GPT won’t make that Bing will.
How do you pay for a pro subscription? The site offers a pricing page, but there's no actual link to subscribe.
There's also no Contact page for sales questions.
Yea I just upgraded my plan to the paid version for $20 a month. Once you sign in, go to this page. claude.ai/upgrade
these are the supported locations. support.anthropic.com/en/articles/8461763-where-can-i-access-claude-ai
@@SkillLeapAI Ok. So we have to sign up first, get the trial version, and then upgrade.
Thanks.
It can't even read links....
Free to use for about 5 prompts then says “You’ve reached the limit…”
Still can’t fully code a simple html page or extract data from a scanned document. I’ll wait on all the gushing 😂
I thought i saw david sach.
Is Bs don't even have real-time internet access
And it still smokes Chat GPT 4 and 4o in coding.
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.
Yea I think ChatGPT is the best for math
@@SkillLeapAI I think the problem is that since these models are basically optimizing for text output and not actually solving or calculating concrete exact solutions like math requires, LLMs are never going to be dependable in this arena. You can be 95% right with a poem, a piece of music, or a video clip and be indistinguishable from human output, but being 95% right about math makes you indistinguishable from Terrence Howard.
This is better but it’s nothing compared to my custom GPT this is gpt 5 level actually with examples why
if anyone wants 10 x vision capabilities for chat GPT i figured out something open Ai forgot about, Its called Smart Vision image/text analysis (Paste your own custom instructions for a superior smart chat)
It is 10 x better at analysing images and especially reasoning relationships read the examples below
I uploaded an image of a cloud that looks like multiple things but it can be interpreted, the one i gave it was a personal photo I took and it has recognised it was a rabbit but not even a random human guessed it every time now on 1st shot so it knows when something is unusual about an image even if you dont say anything, it can also do iq test image reasoning pattern questions relatively well.
Another example is upload an image of a model with clothes that fit or don’t fit or you don’t know it can analyse that and tell you with great detail why it fits or doesn’t fit
It kind of even understands real logic games when giving good instruction but is limited there by the model not the instruction
*IMPORTANT* just gotta follow the instructions given to get the right seed its 1 in 2 chance or so i have absolutely no idea why it needs that. Just paste the conversation starter and you’ll understand what to do.
What is your gpt? Can i access it?
Well... Gemini is smart and cheap enough (in comparison)
Because they’re not as good in my limited opinion. But you have what you need so stay with Gemini
OH my word..... At first test, I found it terrible. Writing an essay. The slang - NO WAY.. Editor score on full essay = 87% with lots of slang words like GONNA. (Going to).
aya:35B running locally = 98% with one spelling mistake. OK it runs slow even running on the GPU but you cannot beat it and nothing will ever beat it in my opinion. It really does blow your mind.
is claude free ? why asking for my phone nomber?K
Too much AI here and there
ollama is better
Private data makes for better responses!
What’s the point of using Claude when it continually refuses to answer the most innocent questions. Too many cautions, useless. If they are so scared to release an AI model, they shouldn't release it
This model might change that
Can you give examples of ur qns ?
I have never had a Claude model refuse to answer and have found it to be very useful.
It’s true, too many caution
I agree. What is even the point? Let it be open
Claude is too slow not competitive with chat gpt fake analysis