Hiya. Sarah here. Aaron’s out today but before he left he wrote about the big new launch from Meta, Lllama 3.1, and why it’s a big deal. But first, I’ve got a few more headlines for you:
📱 Apple is working on a foldable phone that could be released by 2026.
🌦️ Google Researchers built a new weather prediction model using AI.
👾 Elon Musk says Grok 3.0 is coming in December and will be “the most powerful AI in the world.”
Earlier this year, Meta hinted at a significant development in AI: an open-source model that could rival the best from companies like OpenAI. Today, Meta officially launched Llama 3.1, claiming it surpasses other leading models such as GPT-4o and Anthropic’s Claude 3.5 Sonnet in several benchmarks.
Llama 3.1, the largest open-source AI model to date, boasts 405 billion parameters and has shown superior performance compared to its competitors. The Llama-based Meta AI assistant is also expanding to more countries and languages and now includes a feature for generating images based on specific likenesses. CEO Mark Zuckerberg predicts Meta AI will become the most widely used assistant by the end of the year, overtaking ChatGPT.
Meta has been tight-lipped about the data used to train Llama 3.1. The company disclosed that it used synthetic data—generated by a model rather than humans—to enhance the 405-billion parameter version of Llama 3.1 from its smaller 70 billion and 8 billion versions. Ahmad Al-Dahle, Meta’s VP of generative AI, believes Llama 3.1 will be popular with developers as “a teacher for smaller models that are then deployed” in a “more cost-effective way.”
Llama 3.1 is significantly more advanced than its predecessors, having been trained using over 16,000 of Nvidia’s H100 GPUs. Despite the high development costs, Meta is releasing Llama 3.1 with an open-source license, encouraging contributions and modifications from developers. Zuckerberg argues that open-source AI models will outpace proprietary ones, similar to how Linux became the dominant open-source operating system.
In an era where AI is reshaping how businesses operate, the journey of building an early-stage startup has never been more dynamic—or complex. How do founders navigate finding product-market fit, delegation, and scaling, all while adapting to technological innovations?
Join on January 14 at 3 pm PT for a fireside chat with Christina Cacioppo, CEO and Co-founder of Vanta, and Eric Ries, author of The Lean Startup and Founder of LTSE, as they explore the journey of the modern startup founder.
Eric and Christina will discuss:
FOR EVERYONE
Overlap is like email meets Tinder but for networking. It sends highly-relevant, professional matches to your inbox based on industries, skills, and job functions. If you’re both interested, Overlap will make a warm email intro. And it’s all free.
FOR CREATORS
KwaKwa lets creators and professionals create paid courses to share their expertise, fully on mobile. It can be easy as creating an IG story, with stickers, quizzes, polls and more, while KwaKwa takes care of billing, landing pages, support, and delivery.
Tempest AI is an RPG game creator. Enter a single prompt and it automatically generates all your game assets and creates the default game logic. You can dive deeper into customization if you want, and create an infinite RGP game.
FOR BUILDERS
Sendune is a lightweight, open-source HTML email designer. The builder creates pure HTML. Then you can export as a template withour needing any email service providers.
Permit Share-If is a suite of prebuilt, embeddable UI components to make access sharing in applications simple and safe.
