hey sir im working currently on a project.. that will analyse the past 8-10 years of semester exam papers and create a detailed analysis report that will have current trends , important topics etc. will RAG be good or large context window of gemini? actually i want to make it avalaible for all students in my college
Their product marketing is terrible but it is better than open AI which is over hyping their products. At least Google is lowkey just coming out with stuff each week and then tweeting about it.
@@1littlecoder the reason I mention this is Sam/Open AI have been hype machines which is fun, but if you don’t release stuff timely, the internet machine turns on you and that has been happening with Open AI. Granted, now that Open AI have stated this is release season, maybe this will change quickly. Hype without results is terrible. Hype with results is magical. With Google, no hype, no news cycles, but great products speak for themselves. Now they just need an LLM that is better than Open AI and they win (doubt this will happen).
@@JosefTorkelsenThe rolled out a paper a few days ago and it is basically Strawberry but in a broader paradigm of reasoning models, let's hope they will implement it in Gemini.
Seems good but GPT 4o--mini is almost 10 times cheaper for 128k window, it doesn't seem to be much more useful than get 4o-mini, considering the cost benefit analysis.
Ya this is true like when I am using Gemini 1.5 pro from when it launch it is out performing like I am student and facing some issue with physics question jee level question and when I ask question to gemini 1.5 pro it gives me correct answer (not in every question) but ya this is the best model I have ever testes and also in coding but ya there are some issue with the security of the model like the model not generate code sometime becoz of nothing like Idk why but I think laziness ig but ya this is what and flash model is also very fast but not correct as gemini 1.5 pro but when I use same modle in Google gemini website the model works worse like I rarely open gemini website idk why Google made their gemini prompt very bad like not good in coding or other things if Google use their Gemini model directly this will works great but idk what Google is trying to do and ya google just released their new gemini 1.5 flash model and this model is out performing very well I just tested with physics question JEE level and I have total 11 question only 2 answer is in correct otherwise this model is very well.
can you share the link of the blog? and one more question , would you rate gemini 1.5 pro better then gpt4o , in terms of general tasks and video analysis( like OCR , object recognition etc etc) ??
Exactly! Gemini is the worst of the big provider modules. Google should focus on producing a better model. Using gemini just makes want to use claude or chatgpt.
It's still too much for an AI model that is more of a search engine than an AI model at the level of Sonnet. Personally, I only use it because, with many API keys, you can get substantial usage for free. As a user who uses Gemini over 10,000 times a month, I can say that this context is fully utilized by the model only when you point out specific information. On its own, it is impaired in handling a context larger than 8k tokens.
@@1littlecoder sorry, my bad. It was just a rumour. The real news is 👇🏾 “Today, we’re releasing two updated production-ready Gemini models: Gemini-1.5-Pro-002 and Gemini-1.5-Flash-002”
Thanks for always keeping us up to date!
Thank you for your kind support Jim!
love that you get this news out FAST! thanks
Thank you :)
@@TheRealHassan789 that’s not the best part. Making a video and posting it with the same speed is the best one. ✨
Great info! So far I didn't use Gemini, as it wasn't top model in any category. Now it might be more appealing.
Yeah not a lot of people used Gemma!
I think they have used an advanced type of model distillation on an advanced student model
hey sir im working currently on a project.. that will analyse the past 8-10 years of semester exam papers and create a detailed analysis report that will have current trends , important topics etc. will RAG be good or large context window of gemini? actually i want to make it avalaible for all students in my college
Is there a video which shows how much tokens are used for different inputs and output?
Their product marketing is terrible but it is better than open AI which is over hyping their products. At least Google is lowkey just coming out with stuff each week and then tweeting about it.
This is the first comment I've seen where someone said Google's marketing is better than OpenAI's :) Interesting!
@@1littlecoder the reason I mention this is Sam/Open AI have been hype machines which is fun, but if you don’t release stuff timely, the internet machine turns on you and that has been happening with Open AI. Granted, now that Open AI have stated this is release season, maybe this will change quickly. Hype without results is terrible. Hype with results is magical. With Google, no hype, no news cycles, but great products speak for themselves. Now they just need an LLM that is better than Open AI and they win (doubt this will happen).
@@JosefTorkelsenThe rolled out a paper a few days ago and it is basically Strawberry but in a broader paradigm of reasoning models, let's hope they will implement it in Gemini.
@@TheRealUsername please tell me the paper, fellow internet user, i desperately need it, ASAP
@@CoolIcingcake3467 "Chain of Thought Empowers Transformers to Solve Inherently Serial Problems"
Sorry for being late
Seems good but GPT 4o--mini is almost 10 times cheaper for 128k window, it doesn't seem to be much more useful than get 4o-mini, considering the cost benefit analysis.
4o mini is equivalent to Flash not Pro
Lets see if advanced voice from openAI is released today. I heard a report that it would be made more available today.
eyyy they released it like an hour after I made this comment hahaha
Google is underrated
Gemini Flash is literally the cheapest mid-range model right now
Hi please what kind of pc graphicard is needed to handle this. Do you use docker ?
This is not an open model, you need to use it via their api!
Ya this is true like when I am using Gemini 1.5 pro from when it launch it is out performing like I am student and facing some issue with physics question jee level question and when I ask question to gemini 1.5 pro it gives me correct answer (not in every question) but ya this is the best model I have ever testes and also in coding but ya there are some issue with the security of the model like the model not generate code sometime becoz of nothing like Idk why but I think laziness ig but ya this is what and flash model is also very fast but not correct as gemini 1.5 pro but when I use same modle in Google gemini website the model works worse like I rarely open gemini website idk why Google made their gemini prompt very bad like not good in coding or other things if Google use their Gemini model directly this will works great but idk what Google is trying to do and ya google just released their new gemini 1.5 flash model and this model is out performing very well I just tested with physics question JEE level and I have total 11 question only 2 answer is in correct otherwise this model is very well.
can you share the link of the blog? and one more question , would you rate gemini 1.5 pro better then gpt4o , in terms of general tasks and video analysis( like OCR , object recognition etc etc) ??
I use Gemini mainly for two - 1. Long Context, 2. Video and Image -> otherwise I don't find them that great!
💯
OCR and long context Gemini is the leader.
Exactly! Gemini is the worst of the big provider modules. Google should focus on producing a better model. Using gemini just makes want to use claude or chatgpt.
Thanks 🎉
It's still too much for an AI model that is more of a search engine than an AI model at the level of Sonnet. Personally, I only use it because, with many API keys, you can get substantial usage for free. As a user who uses Gemini over 10,000 times a month, I can say that this context is fully utilized by the model only when you point out specific information. On its own, it is impaired in handling a context larger than 8k tokens.
is the rate limit per api key or per account?
🎉
Waoo
To bad it sucks
Gemini 2 is going to be announced tomorrow. 🎉 (sorry fake news)
seriously?
fake news
@@1littlecoder sorry, my bad. It was just a rumour. The real news is 👇🏾
“Today, we’re releasing two updated production-ready Gemini models: Gemini-1.5-Pro-002 and Gemini-1.5-Flash-002”
no worries, that's what I covered here :)
The real news 👇🏾
“Today, we’re releasing two updated production-ready Gemini models: Gemini-1.5-Pro-002 and Gemini-1.5-Flash-002”