I Found a Way to Combine Claude 3 and GPT-4
Вставка
- Опубліковано 6 бер 2024
- Claude 3 has 3 major limitations. It doesn't have up-to-date information or live web access. It can't create images, and it's very limited in file size upload for data analysis.
GPT-4 on the other hand doesn't write as well as Claude 3 and falls short of Claude 3 in several categories.
In this video, I'll share an alternative that allows you to use Claude 3 and GPT-4 in the same place and has access to image generation models such as DALLE.3 and Stable Diffusion XL.
Master ChatGPT, Midjourney, and top 50 AI tools with Our New AI Education Platform.
Start a free trial today: bit.ly/skillleap
Master ChatGPT, Midjourney, and top 50 AI tools with Our New AI Education Platform.
Start a free trial today: bit.ly/skillleap
Great insight on leveraging both Claude 3 and GPT-4's strengths! For entrepreneurs, integrating these tools can streamline workflows. For instance, using GPT-4's nuanced text output for initial drafts and Claude 3's precision for data-driven tasks can create a powerful combo for content and analysis.
This is golden info. Really appreciate it. Perplexity is undervalued.
For an extended period, I have utilized both Perplexity and GPT-4. In my experience, they demonstrate expertise in distinct domains. While I have seen testing results concerning Claude Opus, they do not sufficiently persuade me to transition to this model. However, I anticipate the potential release of GPT-4.5 in the near future.
Oh, dang! I was thinking that switching from Claude pro to this was a no brainer, but the 5 per day limit killed that. I'm loving Claude pro for programming -- even more than GPT4 with a highly optimized custom instructions.
Yea I’m hoping that limitation is temporary.
Now it's unlimited
@@iDrakeOneConfirmed. Thanks!
@@iDrakeOneso they lifted the 5 per day limit?
Great video - give us even more with custom prompts!
Why not use something like OpenRouter or TypingMind instead of Perplexity? Both of these have all the models this video touched on without the context limit?
Really liked this video, shared it!
Is there a way to force Perplexity to use Sonnet so I won't waste my 5 Opus searches? Also, I would really like to see a comparison of Sonnet to GPT4, and to Opus, to see how this free models fares against the best. Thanks!
for some reason, it only has one claude 3 setting. Good idea on the comparison video. I'll add it to the list.
The problem I had with perplexity is the context. You will ask something, it answer, you ask a follow up question and sometime it just answer it if it was a brand new conversation. I guess they use very small context window to save on api cost. Also Dall-e is 1 images at a time (Chatgpt is 2 and Copilot pro is 4).
yea good point. Which LLM do you use for perplexity?
@@SkillLeapAI I was using it with GPT4, but I since cancelled. Now I use mostly only Cursor for programming and you can RAG your own documentation, which make GPT4 output really on point. It's quite impressive when the RAG of the documentation is part of the context the difference in quality.
Does the write mode solve it?
The Glibatree GPT gives you 4 images generated, and you get the midjourney code generated as well to copy and paste
I just had the same issue, have you tried asking the model yourself?
My question is, how does this work with subscriptions (or lack thereof) to one of these LLMs that it uses. IE, does Perplexity allow 5 Opus questions per day even if you don't have an Anthropic paid account? What if you do have an Anthropic paid account, does it still limit you to 5 per day even so? What about any similar limitations to GPT4?
It doesn’t connect to your other accounts. So it doesn’t change if you have GPT 4 account or Claude pro account. you get 5 chats with Claude opus which I’m hoping they change soon. And only GPT4 limit is if you hit your total message limit I believe, which is 600 per day
can we use Sonnet instead of Opus?
Wait so do you get the model You get in chatgpt plus but with 600 messages instead of 40? Or is there a catch.
The catch is the context window that is really small.
I know Perplexity is great for research but how is it for content creation? Copilot can access the internet while it writes, so can Gemini. How would you compare Perplexity in regards to content creation with those three?
I prefer GPT4 for content creation but I’ll test it out now with Claude 3 inside perplexity and see if it’s better
Thanks!!@@SkillLeapAI
i wonder how perplexity is able to bring claude 3, and gpt4 for free of cost,. any thoughts? if there is 0$ spend on gpt4, mixtral, claude 3 like models within perplexity, for the cost of 1 (which we pay to perplexity) where its also bringing in SDXL, Dall 3 (assuming they are also free), wondering how perplexity is able to provide all such at 20,. ofcourse users wish to see more for that price, but its a good starting point and the more users the better pricing reductions down the line ;) love to hear more comments or inputs on the above.
I’m guessing they’ve done the math and they still make a profit after they pay OpenAI and Anthropic their API fees. And they have the daily usage cap of 600 across all answers to make sure someone can’t be a super power user.
I use five different AI programs: Gemini, chatGPT4, Perplexity, Claude, and Groq. Most of them paid versions. I agree with you, Perplexity is the best as of now.
I think Gpt is still the mother honsetly speaking.
@@Joseph-nw3gw lazy on coding and math
@@Planthier76 What would you recommend for coding?
@@xanderunderwoods3363 looks like Claude 3 opus is really nice at it. But not avalaible in Europe ...
I use perplexity since few days, it looks amazing for searching infos but messy at working like a traditional GPT.
If you pay for Perplexity Pro, do you also have to pay for ChatGPT4, Claude Opus, Dalle3, Stable Diffusion, etc.
Also want to know
From $20Proplexity Pro, use can choose any model without paying any other money. Go to Setting to choose any model
Is anyone else experiencing really low prompt limits with Claude 3? I have the paid version and swear it only lets me do like 15 prompts a day
Guys whould you use perplexity or chatgpt Plus?
Its hard to choose one. I use both. I can’t work without customGPTs now, so I need GPT plus
It also isn’t aware there is a Claude 3 opus to subscribe to
How much does perplexity charge per month
The paid option is 20 a month
🥳👏👏
There was an offer… if you buy a rabbit 🐰 one you get a free year of perplexity pro at the same price so it’s basically a free rabbit.
Now that the 5 opus message limit has been removed, I upgraded to Pro. I experimented with coding for a bit (not extensive) and have not found Perplexity with Opus to be as good as native Claude Opus. In Claude, it followed instructions very well and output code that worked the first time. I clarified some of the features and asked for new ones and every time it created working code and mostly interpreted my intent well.
Claude Opus via Perplexity was completely different. It's first attempt wouldn't even run. Even after several rounds of explaining what it did wrong it failed to produce anything close to the quality that native Claude did. It even forgot that it had written code and asked me to provide the faulty code so it could correct it.
So far, I can't see myself using Perplexity for coding. ☹
Why do u think this happens? It’s the same model afterall
@@ahmadalissa9187they probably limit the models hence the pricing ? I heard on Reddit that the context window was smaller (30k token). But all of this is speculation as they are not transparent on this. But I have the same feeling that the models are tamed down a bit. Depending on your needs could be enough or maybe not.
It's a neat concept but it's misleading to say its combing them. It does not remember conversations either. This is a side by side, not a combination. It's just an AI asset library.
I ❤Perplexity, just disappointed they don’t have Claude 3 Opus in their API yet… not sure if this is coming or not, would be nice to have…
everyone can skip this. this is an adv like promoting poe, pkexity
Will do, thank you!