• AI Academy
  • Posts
  • 🤖 Meta Launches GPT-4 Competitor for Free

🤖 Meta Launches GPT-4 Competitor for Free

Plus: Meta skips EU release and OpenAI debuts cheapest model

July 25, 2024

Hello AI Enthusiast,

Just another week full of AI releases. Meta introduced a new open-source AI model that competes with top models. However, there are challenges with regulations in Europe. At the same time, OpenAI released a new, less expensive AI model. These releases could affect how companies and people use AI now and in the future.

Read on to learn more, but before you do, we have two upcoming free webinars next week that might interest you.

📅 July 30 at 5 PM CEST - Join our CEO Gianluca Mauro together with Innovation Strategist John Vetan in the "Start with AI, the Right Way" webinar, where they'll provide a framework for executives to understand and confidently launch AI projects. Register here.

📅 July 31 at 6 PM CEST - Gianluca Mauro will provide you with a practical framework to help organizations understand and leverage the impact of AI. Register here.

The Big Picture 🔊

Meta Launches Llama 3.1 Open Source AI Model

Meta has launched Llama 3.1, its largest open-source AI model yet. They're claiming it matches top models but runs at half the cost of OpenAI's GPT-4. Meta's not going solo on this one; they've partnered with cloud providers like AWS and Azure to make Llama 3.1 widely available. They’ll roll it out first on WhatsApp, with plans to expand to Instagram and Facebook down the line.

Gianluca Mauro
Gianluca MauroCEO and AI Rockstar

OpenAI offers its proprietary model exclusively, while Meta is making Llama 3.1 available on multiple cloud providers. These platforms provide the infrastructure to deploy and run AI at scale. Llama 3.1 is a free, "partially” open-source model, customizable but with hidden internals.

As implementation becomes easier, companies seeking to customize services and run operations on their servers will likely adopt models like Llama. Using cloud platforms for this also provides a responsible party for support.

For a deeper take on LLama 3.1, Gianluca shared a TikTok. 🤳

@gianluca.mauro

Meta democratizing, intelligence #Ai #learnontiktok #artificialintelligence #business #machinelearning #product #ux #entrepreneur

Meta Blocks AI Models in EU

All exciting for users and companies, except those in Europe who won't get to try it since Meta has announced it will not release its AI models in the EU due to regulatory uncertainty. This decision underscores the growing tensions between U.S. tech firms and EU regulators, as Meta faces challenges complying with Europe's strict data protection laws.

Helin Yontar
Helin YontarCPO and Podcast Host

Meta is being cautious with EU rules, following Apple's recent example. They're treading carefully to avoid potential penalties. But here's the thing: Big Techs and Europe need each other. It's all part of a big negotiation game.

We think these tech giants are hoping to get the public on their side to nudge the EU towards more flexible rules.

OpenAI Launches Affordable GPT-4o mini Model

Llama 3.1 wasn't the only new AI model announced recently. OpenAI has introduced GPT-4o mini, a highly affordable AI model priced at 15 cents per million input tokens and 60 cents for output tokens, making it over 60% cheaper than GPT-3.5 Turbo, yet just as fast. GPT-4o mini is now OpenAI’s most economical model to date. It is multimodal, accepting both text and image inputs and producing text outputs, and is intended for smaller tasks, including vision tasks.

Gioele Mottarlini
Gioele MottarliniCOO and Image Addict

The introduction of ChatGPT-4o mini is another indicator that AI is progressing towards commodity status. API costs have dropped, and OpenAI is encouraging developers to transition from ChatGPT-3.5 Turbo to 4o mini.

This intensifying competition between tech companies is driving prices down, benefiting users, developers, and companies utilizing AI technology.

Is the AI Price War Good for the Industry?

AI model prices are dropping due to fierce competition. Is this trend towards more affordable AI beneficial for the industry overall?

Login or Subscribe to participate in polls.

Bits and Bobs 🗞️

  • Andrej Karpathy, former head of AI at Tesla and OpenAI researcher, has launched Eureka Labs, an "AI native" education platform aimed at creating AI teaching assistants

  • OpenAI has introduced new tools for ChatGPT Enterprise to enhance compliance, data security, and user access.

  • Major tech companies including Google, OpenAI, Microsoft, and Amazon have joined forces to form the Coalition for Secure AI (CoSAI) to develop best practices for AI security.

  • Mistral AI and NVIDIA have released the Mistral NeMo language model, combining Mistral’s training data expertise with NVIDIA's optimized hardware.

  • Mistral introduced two new AI models, Codestral Mamba and Mathstral, which promise faster inference times and better performance for coding and math reasoning tasks, respectively.

  • Google announced Project Oscar at Google I/O Bengaluru, an open-source platform enabling AI agents to manage software programs by monitoring bugs and facilitating the development lifecycle.

  • Figma temporarily removed its "Make Designs" generative AI tool after finding it was producing designs too similar to real-world applications.

  • Microsoft researchers have developed SpreadsheetLLM and SheetCompressor to help LLMs analyze and manage spreadsheet data.

  • Miro has unveiled the Intelligent Canvas, a new AI-enhanced platform designed to streamline collaboration.

Tribal News🫂

So, a member of our community asked suggestions regarding what are the best AI tools to create presentations and apparently another member, 4 days later, built a prototype. Wild! 🤯

LOLgorithms 😂

Some people are eagerly awaiting GPT-5.

That's a wrap on our newsletter! Here’s a quick recap before you go:

Catch you next week! 👋