Crazy how these AI models are upgrading so fast! Which one are you using the MOST now? If you enjoyed this video, I invite you to also join my digital growth newsletter: www.graceleung.com/newsletter/, see you inside! 🙂
@@graceleungyl well i regularly make PDF documents from Pictures i take from Notes of pages (screenshots acutally ) . and when i upload the pdf document of images, the chatgpt 4 o version says that the contents of the pdf document are empty and it cant read the pages . but when i upload the very same pdf document , perplexity can actually read and answer the questions from the very same pdf document . it clearly shows that Perplexity is on another level .
@@graceleungyl I signed up for ChatGPT Plus a few weeks ago and have been using it every day for various things. Extremely useful, I'm starting to look at it like the Internet itself, I can't imagine going back to not having it. Not sure how it compares to the others, but for my uses it doesn't matter too much. It's much better than nothing, and it's easily worth the $20/mo for the latest updates and no limits.
Thanks for the video! In your opinion, which model is better for working on long coding projects that involve uploading large files and frequently referencing them during the conversation: OpenAI o1-preview or Claude Sonnet 3.5? Does Claude still have a chat size limitation in the pro version? Does the "Projects" feature help mitigate this limitation by allowing files to be referenced without reuploading them each time? How does the o1-preview model handle this scenario? Does it have a chat size limit in the pro version?
Halfway through the video and just needed to comment how well this video was made. I am currently looking for a job after college and only have about $20 dollars to spend on one of these models and this was exactly the video I was looking for. I will definitely be checking out your other videos Grace this is amazing!
Just discovered this channel, and I have to say, the videos are incredibly detailed and practical-unlike many others that just offer simple and superficial overviews
I 100% think Claude and GPT together make the powerhouse most people would need. I’ve never really used perplexity for anything other than research so it was interesting to see it do these kinds of tasks
Agree, and to add on that, I found Claude project is also super useful for specific tasks project with knowledge... For Perplexity, most people is skeptical if they can treat it as an "all-in-one" AI because of multiple AI models, so part of the video is to answer that as well.
Gotta say, GPT paid is (as of today) the best one to be used in education. Which is a bummer because I genuinely like the fact that Gemini knows quite literally everything about my teaching style and has access to all of my slides. However, many, many times Gemini simply generated the wrong information. Wrong book chapter's description, wrong pages of books, wrong numbers when it comes to finding information and, let's be honest, not the best tone when you want to send emails if compared to GPT. Also, the new language model is a big hit around kids. Google, I love you and all but please make Gemini at least half of what gpt currently is.
Thanks adding that! Yah... exactly what I found when using Gemini (or even advanced), it's not bad and creative in certain tasks, but definitely is expected to be more outstanding with such amount of resources invested from Google. Still I believe it will keep improving, particularly I hope they can add real-time search. It's gonna be a big deal.
I really hope they catch up. I've been a paid member of chatgpt for about a year now and other than a few occasions haven't played with Gemini because it doesn't seem to work as well for me. It's a shame, because all of my electronics (home and phone) are Google and I bought into them in anticipation of using a cohesive ecosystem. Was really hoping for AR glasses because they bought focals by North, but that disappeared never to be heard from again a few years ago just as I was going to pre-order. Come on Google, get this stuff out for us, focals already had a dedicated user base and now it's crickets.
I can feel you 😢 To be fair, I don't think Google AI tech is dumb, it's indeed powerful (take a look at NotebookLM). My guess is they have been very cautious which can be a good thing, but when done too much... well...
@@graceleungyl your guess isn't even half right. It's not that Google is too careful; it's that they're asleep at the wheel. The interview with former Google CEO Eric Schmidt at Stanford is revealing: Google employees aren't anymore driven, unlike engineers at OpenAI and Perplexity, both of which are seeking huge funding rounds. The criticism of Gemini is all-around bad. I tried it with coding, verifying historical facts, re-writing medical literature, summarizing a PDF. In every task, Gemini was beaten by either ChatGPT or Perplexity
hi grace, which is the best AI tools to transcribe english and mandarin audio to words , in the past i am using whisper jax (free) but very unstable , any recommendation ?
Are you looking to transcribe and then translate? I would actually recommend you first use Notebooklm to get the transcript (just upload the audio file), then you can use Claude to do the translation. I would suggest trying Claude first for this case, as it's more attention to details. And depending on your audio length, you may need to ask it to transcribe one section by section to maintain the context.
I rarely use them for code like SQL, so can't comment 100% accurately, but GPT4 is more widely recognized in its strong coding ability, you can try GPT first. As for websearch, at this moment, Perplexity Pro still my go-to option. Though, it may be different as SearchGPT is fully rolled out, we'll see.
There's a difference between the native Claude and accessing it on Perplexity. If you just use it for simple writing stuff, then just getting Perplexity should be fine. But if you need some other features like artifact generation, setting project base, then Claude is still preferred in my opinion.
I am a full stack developer and i want to super charge my productivity , i am very confused shoud i get claud pro or gpt 4 pro , i like claud but the chat length limit sucks!
I know, that's a headache. If you need to do lots of coding task, perhaps you can check ChatGPT first, they now have the Canvas mode, then you can combine the o1 thinking model and use Canvas to build code and ask feedback. And you can compare it with Claude's output see which one is better.
nice video 👍🏻. i want to ask from which platform can I give the RUclips video URL so that the AI model can summarize the videos , if any modell can do that at all. and how did you give the models the RUclips videos, transcript and comments transcript; can you tell me how you made that? i would be very thankful to you
Thanks! I used an online free RUclips transcript generator, I have put the link in the description. For comments, you can just copy them all from the video watch page, and then paste them directly on Claude (It will be automatically turned as a attachment)
Thanks for your input, yah I agree, always effective to break a complex prompt into small tasks, just because I need to use the same prompt for evaluation and see how they all response.
Not trying to troll. Replay your video at 0.75 speed. If you speak around that speed, your videos will be easier for the English-speaking to understand
Interesting. I played it at 1.75x then moved it to 1.5. Works well at this speed, too, which is an exact 2x multiple of your recommended .75. I think you’re onto something here. Now if we could only select 2.25 and try that out. Hmm
One way I do is setting custom instruction in project or you instruct it in the prompt and ask it to avoid any verbiage or unnecessary explanation in the output. That will help to avoiding lengthy output that takes up usage.
Crazy how these AI models are upgrading so fast! Which one are you using the MOST now?
If you enjoyed this video, I invite you to also join my digital growth newsletter: www.graceleung.com/newsletter/, see you inside! 🙂
ChatGpt plus but after seeing your video i think i might switch over to Perplexity pro 😂
😂 But I'm sure GPT will keep adding new features (same as Perplexity Pro), so it's still a difficult decision to make LOL
@@graceleungyl well i regularly make PDF documents from Pictures i take from Notes of pages (screenshots acutally ) . and when i upload the pdf document of images, the chatgpt 4 o version says that the contents of the pdf document are empty and it cant read the pages . but when i upload the very same pdf document , perplexity can actually read and answer the questions from the very same pdf document . it clearly shows that Perplexity is on another level .
@@graceleungyl I signed up for ChatGPT Plus a few weeks ago and have been using it every day for various things. Extremely useful, I'm starting to look at it like the Internet itself, I can't imagine going back to not having it. Not sure how it compares to the others, but for my uses it doesn't matter too much. It's much better than nothing, and it's easily worth the $20/mo for the latest updates and no limits.
Thanks for the video! In your opinion, which model is better for working on long coding projects that involve uploading large files and frequently referencing them during the conversation: OpenAI o1-preview or Claude Sonnet 3.5?
Does Claude still have a chat size limitation in the pro version? Does the "Projects" feature help mitigate this limitation by allowing files to be referenced without reuploading them each time?
How does the o1-preview model handle this scenario? Does it have a chat size limit in the pro version?
Halfway through the video and just needed to comment how well this video was made. I am currently looking for a job after college and only have about $20 dollars to spend on one of these models and this was exactly the video I was looking for. I will definitely be checking out your other videos Grace this is amazing!
Thank you! So glad to hear that, hope you find your ideal upgrade option with some directions to start!
What did you decide to go with ?
Just discovered this channel, and I have to say, the videos are incredibly detailed and practical-unlike many others that just offer simple and superficial overviews
OMG.. that means a lot, thank you for your kind words!
I 100% think Claude and GPT together make the powerhouse most people would need.
I’ve never really used perplexity for anything other than research so it was interesting to see it do these kinds of tasks
Agree, and to add on that, I found Claude project is also super useful for specific tasks project with knowledge... For Perplexity, most people is skeptical if they can treat it as an "all-in-one" AI because of multiple AI models, so part of the video is to answer that as well.
Definitely the best video on this burning topic !!
Which is best in accuracy ma'am pls let me know thankyou ❤🌟
You're doing great work!
Thank you!
Gotta say, GPT paid is (as of today) the best one to be used in education. Which is a bummer because I genuinely like the fact that Gemini knows quite literally everything about my teaching style and has access to all of my slides. However, many, many times Gemini simply generated the wrong information. Wrong book chapter's description, wrong pages of books, wrong numbers when it comes to finding information and, let's be honest, not the best tone when you want to send emails if compared to GPT. Also, the new language model is a big hit around kids. Google, I love you and all but please make Gemini at least half of what gpt currently is.
Thanks adding that! Yah... exactly what I found when using Gemini (or even advanced), it's not bad and creative in certain tasks, but definitely is expected to be more outstanding with such amount of resources invested from Google. Still I believe it will keep improving, particularly I hope they can add real-time search. It's gonna be a big deal.
I really hope they catch up.
I've been a paid member of chatgpt for about a year now and other than a few occasions haven't played with Gemini because it doesn't seem to work as well for me.
It's a shame, because all of my electronics (home and phone) are Google and I bought into them in anticipation of using a cohesive ecosystem.
Was really hoping for AR glasses because they bought focals by North, but that disappeared never to be heard from again a few years ago just as I was going to pre-order.
Come on Google, get this stuff out for us, focals already had a dedicated user base and now it's crickets.
I can feel you 😢 To be fair, I don't think Google AI tech is dumb, it's indeed powerful (take a look at NotebookLM). My guess is they have been very cautious which can be a good thing, but when done too much... well...
@@graceleungyl your guess isn't even half right. It's not that Google is too careful; it's that they're asleep at the wheel. The interview with former Google CEO Eric Schmidt at Stanford is revealing: Google employees aren't anymore driven, unlike engineers at OpenAI and Perplexity, both of which are seeking huge funding rounds. The criticism of Gemini is all-around bad. I tried it with coding, verifying historical facts, re-writing medical literature, summarizing a PDF. In every task, Gemini was beaten by either ChatGPT or Perplexity
These were solid comparisons! Thanks for sharing, Gracie! 😎
Thanks Eddy!! 😁 Loving your channel as well! :))
@@graceleungyl Thank you 😎
hi grace, which is the best AI tools to transcribe english and mandarin audio to words , in the past i am using whisper jax (free) but very unstable , any recommendation ?
Are you looking to transcribe and then translate? I would actually recommend you first use Notebooklm to get the transcript (just upload the audio file), then you can use Claude to do the translation. I would suggest trying Claude first for this case, as it's more attention to details. And depending on your audio length, you may need to ask it to transcribe one section by section to maintain the context.
@@graceleungyl thanks grace , i shall give it a try and let you know
Hiii how much cost all the packages that are you using?
Yeah. It really depends on what you're doing. Claude and chatGPT are usually my go-to for serious work. I use metaAI and Gemini to check ideas
O yah, we still have Meta AI! Using a mix of tools is indeed a smart way to maximize the value. Thanks sharing your experience!
on which platform do you Use metaAI ? i tried to find a iOS App but couldnt find in App store
@@graceleungyl 💯
Which is the best for SQL and websearch? I can't decide between ChatGpt plus and perplexity pro. Great video!
I rarely use them for code like SQL, so can't comment 100% accurately, but GPT4 is more widely recognized in its strong coding ability, you can try GPT first. As for websearch, at this moment, Perplexity Pro still my go-to option. Though, it may be different as SearchGPT is fully rolled out, we'll see.
Is Perplexity using the Claude model? If so why would you need a paid subscription to both?
There's a difference between the native Claude and accessing it on Perplexity. If you just use it for simple writing stuff, then just getting Perplexity should be fine. But if you need some other features like artifact generation, setting project base, then Claude is still preferred in my opinion.
How about copilot, compared to Chatgpt. Would you recommend?
Planning to a video about Copilot, what do you want to hear about?
Finally, a real world scenario than "Fruits" hahah
We've all seen enough 😂
@@graceleungyl say no to fruits hahah
Thank you for providing these useful comparisons.
Thanks Victor, glad it helps!
I am a full stack developer and i want to super charge my productivity , i am very confused shoud i get claud pro or gpt 4 pro , i like claud but the chat length limit sucks!
I know, that's a headache. If you need to do lots of coding task, perhaps you can check ChatGPT first, they now have the Canvas mode, then you can combine the o1 thinking model and use Canvas to build code and ask feedback. And you can compare it with Claude's output see which one is better.
Great comparison 👍
Thanks for watching!
Hey recent addicted to your videos❤
Thanks for support!
nice video 👍🏻. i want to ask from which platform can I give the RUclips video URL so that the AI model can summarize the videos , if any modell can do that at all. and how did you give the models the RUclips videos, transcript and comments transcript; can you tell me how you made that? i would be very thankful to you
Thanks! I used an online free RUclips transcript generator, I have put the link in the description. For comments, you can just copy them all from the video watch page, and then paste them directly on Claude (It will be automatically turned as a attachment)
@@graceleungyl Thanks. my next question would be which AI does this "Online RUclips Transcript Generator " use 😅
As far as I know, this tool doesn’t use AI to generate the script, it just retrieves the existing transcript from RUclips.
Prompt n 2 is tool long, you need to split it in prts with Chatgpt O1
Thanks for your input, yah I agree, always effective to break a complex prompt into small tasks, just because I need to use the same prompt for evaluation and see how they all response.
10:09 How can I do that? Just tell me what I need to learn. Thank you for your helpful videos
No problem! I used an online free RUclips transcript generator, I have put the link in the description.
For comments, I just copy them from the RUclips watch page, works perfectly
And copilot?
Planning to do a video about Copilot, anything you want to know more in particular?
Not trying to troll. Replay your video at 0.75 speed. If you speak around that speed, your videos will be easier for the English-speaking to understand
I appreciate your feedback! :)
Interesting. I played it at 1.75x then moved it to 1.5. Works well at this speed, too, which is an exact 2x multiple of your recommended .75. I think you’re onto something here. Now if we could only select 2.25 and try that out. Hmm
Claude's new AI Agent for your computer 👀
Haha yes, things keep changing terribly fast 😆
@@graceleungyl I love it. Plus it gives you another excuse to crank out another video :)
How can we take Claude seriously with all those data limits?
One way I do is setting custom instruction in project or you instruct it in the prompt and ask it to avoid any verbiage or unnecessary explanation in the output. That will help to avoiding lengthy output that takes up usage.
@@graceleungyl I meant that you can't upload large files. I wanted to analyze a 35 MB PDF and it wouldn't let me.
@@jake13122 I know.. same here, that's really annoying, I still can't find a workaround
Is this channel with an AI Avatar?
Nope... 😢 100% real human