Open-source AI gets a big boost with Meta's latest Llama models
The goal is to accelerate AI innovation and democratize access to this cutting-edge technology.
Until now, open-source large language models (LLMs) have generally lagged behind their closed-source counterparts in terms of capabilities and performance. That's why Meta is publicly releasing the latest version of its AI models to usher in a new era where open-source will take the lead.
Meet Llama 3.1 405B, which Meta describes as the world's largest and most capable openly available foundation model, as it focuses on collaboration and accessibility.
With over 300 million total downloads of all Llama versions so far, including Llama 3.1 70B and Llama 3.1 8B, Llama 3.1 405B (boasting a staggering 405 billion parameters, a metric that signifies the model's capacity for complex tasks) comes with powerful capabilities in general knowledge, steerability, math, tool use, and multilingual translation, that rivals the top AI models available.
It can expand context length to 128K, and supports eight languages. Using its multi-lingual feature, you can write a prompt for it to translate the story of the Goose that Laid Golden Eggs into Spanish. For software development, you can write a prompt that lets it create a program that generates a perfect maze, using a recursive backtracking algorithm or a depth-first search algorithm, with customizable size and complexity.
Which countries is Meta AI available?
As of July 23, 2024, Meta AI is available in 22 countries, including the United States, Australia, Canada, Ghana, Jamaica, Malawi, New Zealand, Nigeria, Pakistan, Singapore, South Africa, Uganda, Zambia, Zimbabwe, Argentina, Chile, Colombia, Ecuador, Mexico, Peru and Cameroon.
Meta AI was available for a short period in the United Kingdom, but Meta stopped its roll-out in the European Union (EU) region due to regulatory concerns.
You can also interact with Meta AI across WhatsApp, Instagram, Messenger and Facebook in languages including English, French, German, Hindi, Hindi-Romanized Script, Italian, Portuguese and Spanish.
By making Llama 3.1 open-source, Meta is essentially inviting developers around the world to tinker, improve, and build upon the powerful AI tool, something that OpenAI's ChatGPT or Google Gemini lacks.
It remains to be seen how Microsoft, Google and OpenAI will respond, although Elon Musk's xAI has since released its 314 billion parameter Mixture-of-Experts AI model, Grok-1 to the public, earlier this year in March.
One thing is certain: the race to develop the most powerful and accessible AI is heating up, which is great for the future of AI since it has the potential to accelerate AI innovation and democratize access to this cutting-edge technology.
