top of page
  • tech360.tv
    • Apr 20
    • 3 min read

Meta Unveils Llama 3 Large Language Model & Real-Time Image Generator to Compete with OpenAI: Meta AI's Evolution and Strategic Partnerships

Meta Platforms releases early versions of Llama 3 AI model and real-time image generator. Meta AI assistant to receive more prominent billing within Meta's apps and a new standalone website. Partnership with Google to include real-time search results in Meta AI's responses.

Meta Platforms, the social media giant, has unveiled early versions of its latest large language model, Llama 3. Alongside Llama 3, Meta also introduced an image generator that updates pictures in real time while users type prompts. This move comes as Meta aims to catch up to generative AI market leader OpenAI.


The Llama 3 models will be integrated into Meta's virtual assistant, Meta AI, which the company claims is the most sophisticated among its free-to-use counterparts. Meta compared the performance of its AI assistant in areas such as reasoning, coding, and creative writing against offerings from rivals like Google and Mistral AI, a French startup.


To increase the visibility of Meta AI, the updated assistant will receive more prominent billing within Meta's Facebook, Instagram, WhatsApp, and Messenger apps. Additionally, Meta has launched a standalone website that positions Meta AI to compete directly with OpenAI's popular ChatGPT.


The new website greets visitors with various prompts, inviting them to try out the assistant's capabilities. Users can ask the assistant to create a vacation packing list, play 1990s music trivia, provide homework help, or even paint pictures of the New York City skyline.


In an effort to challenge OpenAI's dominance, Meta has been working tirelessly to release generative AI products to its billions of users. This endeavor has involved a costly overhaul of computing infrastructure and the consolidation of research and product teams.


Meta has been openly sharing its Llama models with developers, allowing them to build AI apps. By offering a powerful free option, Meta aims to disrupt rivals' plans to monetise their proprietary technology. However, this strategy has raised concerns about safety and the potential misuse of the models by malicious actors.


Llama 3 has been equipped with new computer coding capabilities and has been trained on both images and text. However, for now, the model will only output text. Chris Cox, Meta's Chief Product Officer, mentioned that future versions of Llama 3 will include more advanced reasoning abilities and the capability to generate both text and images.

Cox emphasised that the ultimate goal of Meta AI is to make users' lives easier by assisting with various tasks, such as interacting with businesses, writing, and trip planning.


In addition to the Llama 3 release, Meta announced a partnership with Google to incorporate real-time search results into the assistant's responses. This collaboration supplements Meta's existing arrangement with Microsoft's Bing search engine.


With this update, the Meta AI assistant will expand to more than a dozen markets outside the U.S., including Australia, Canada, Singapore, Nigeria, and Pakistan. However, Meta is still working on the best approach to launch in Europe, where stricter privacy rules and the forthcoming AI Act pose challenges.


Mark Zuckerberg, CEO of Meta, acknowledged the competition with OpenAI in a video accompanying the announcement. He touted Meta AI as "the most intelligent AI assistant that you can freely use." Zuckerberg also highlighted the favorable performance of the smaller versions of Llama 3, which have 8 billion and 70 billion parameters, respectively, compared to other free models.


While the results are impressive, industry experts note a growing performance gap between free and proprietary models. Nathan Benaich, founder of AI-focused venture firm Air Street Capital, expressed that the results indicate the need for improvement in free models.


Meta addressed previous issues with context understanding in Llama 2 by using high-quality data to train Llama 3. The company did not provide specific details about the datasets used but mentioned that seven times more data was fed into Llama 3 compared to its predecessor.

 

  • Meta Platforms releases early versions of Llama 3 AI model and real-time image generator

  • Meta AI assistant to receive more prominent billing within Meta's apps and a new standalone website

  • Partnership with Google to include real-time search results in Meta AI's responses


Source: REUTERS

As technology advances and has a greater impact on our lives than ever before, being informed is the only way to keep up.  Through our product reviews and news articles, we want to be able to aid our readers in doing so. All of our reviews are carefully written, offer unique insights and critiques, and provide trustworthy recommendations. Our news stories are sourced from trustworthy sources, fact-checked by our team, and presented with the help of AI to make them easier to comprehend for our readers. If you notice any errors in our product reviews or news stories, please email us at editorial@tech360.tv.  Your input will be important in ensuring that our articles are accurate for all of our readers.

bottom of page