Meta has announced Llama 3, the next version of its highly popular open-source large-language model.
The war of AI chatbots has many contenders and to some extent, we couldn’t technically say that Meta was it. The company’s Llama family of models was used internally with no chatbot-like interface. The LLM itself was open-source (Llama 2 being one of the best open-source models to date) and one could set it up for personal, chatbot-like use, but there was no official app to compete with the likes of ChatGPT, Gemini (earlier Bard), Poe, Claude, Bing AI, Perplexity, etc.
But Llama 3 has major text-generation capabilities. According to the official page, the 70B version will outperform Gemini 1.5 and Claude 3 Sonnet. It will be released later this year. The model released so far is the 8B weight which outperforms Gemma 7B and Mistral 7B.
The model card on the GitHub page mentions that “The core values of Llama 3 are openness, inclusivity and helpfulness. It is meant to serve everyone, and to work for a wide range of use cases.” It can be said that this is what Meta has been aiming for since the beginning.
The AI assistant from Meta is now being incorporated into the search of all of its major products like Facebook, Instagram, Messenger, and WhatsApp.
So, the CEOs might not have made it official yet, but everybody knows that the head-to-head battle against AI chatbots starts now. The Verge’s report on Llama 3 is literally titled, “Meta’s battle with ChatGPT begins now.”
If you wanted more proof that Meta is finally in this battle, the company has finally allowed users to interact with the model on a dedicated domain (not available in all countries) – Meta.ai.
