Llama
Meta's open-source family of LLMs
Start new thread
Diego Guerra Orozco

Llama 3.3 70B - Llama 405B-level performance, at a fraction of the cost

by•
Meta’s Llama 3.3 multilingual LLM is an instruction tuned generative model in 70B that for some applications approaches the performance of Llama 3.1 405B and other frontier-level closed source models

Add a comment

Replies

Best
Johannes Danielmeyer
Ohhh - Great job Meta. I love Llama 3.3 ; First we only used you as a fallback, but now as a first option. Thank you for continuing your open source approach and making LLM requests a commodity.
Diego Guerra Orozco
Hunter
@johannesdanielmeyer So glad to hear that Johannes! Would love to understand more about what's working/whats not/feature requests for your use case (the more details the better!)
mhkthink
i use llama3 on addgraph.com/ai, Good results
Diego Guerra Orozco
Hunter
@mhkthink thanks for sharing! Tell me more, what is your use case? any feature requests?
mhkthink
@diego_guerra_orozco1 to generate roadmaps,
Pathange Balaji Rao
LLMs are something that I wish to learn, for the first time something in LLMs had made me go through the documentation. Great work @raghothammurthy @daltonf @ragavan . Keep killing it
Diego Guerra Orozco
So glad to hear that Johannes! Would love to understand more about what's working/whats not/feature requests Would love to make it more useful to you
Jonathan Boutelle
INSANE that we can run real LLMs on laptops now.
William Jin
This could significantly lower my AI costs while maintaining performance. What unique applications do you think it excels in?