- The Rundown AI
- Posts
- Meta's GPT-4 competitor is coming
Sign Up | Advertise | Tools | AI University
Welcome, AI enthusiasts.
Meta is gearing up to give the world a taste of its next-gen AI, with two scaled-down versions of its hotly-anticipated new model set to release within the next month.
But as competitors continue to raise the bar, the question remains: Will Llama 3 be a game-changer, or just another player? Let’s explore…
In today’s AI rundown:
Meta confirms Llama 3 coming within the month
AI chip wars heat up with Intel Gaudi 3
Make presentations 10x faster with AI
Cohere’s Command R+ climbs leaderboard
6 new AI tools & 4 new AI jobs
More AI & tech news
Read time: 4 minutes
LATEST DEVELOPMENTS
META
Image source: Midjourney
The Rundown: At an event in London, Meta confirmed that it plans an initial release of Llama 3, the next generation of its large language model, within the next month.
The details:
Meta will release several different versions of Llama 3 with different capabilities over the year — with the initial release within the next month.
The company did not disclose the size of the parameters used in Llama 3, but it's expected to have about 140 billion parameters and rival OpenAI’s GPT-4.
Meta previously released Llama 2 in three sizes, with the largest at 70B parameters.
Meta has stockpiled 350,000 coveted H100 GPUs over the last year to ramp up AI infrastructure, a supply that dwarfs other competitors.
Why it matters: Meta's confirmation of Llama 3 shows its commitment to catching up to OpenAI. But the open model simply ‘rivaling’ GPT-4 is suddenly not as impressive of a feat (see today’s AI research section). Regardless, open-source AI is on the rise.
TOGETHER WITH AE STUDIO
The Rundown: AE Studio tackles the toughest challenges in tech, like AI alignment. While their R&D team explores the frontiers of AI safety and consciousness, their consulting team delivers cutting-edge AI solutions.
Partner with AE Studio for:
A proven track record in AI, blockchain, BCI, and more
End-to-end expertise, from ideation to deployment
A team that treats each project like their own startup
AE Studio is ready to listen to any big idea and make it a reality. Share your business challenge today.
INTEL
Image source: Intel
The Rundown: Intel just unveiled its Gaudi 3 AI chip at the company’s Vision event, taking direct aim at Nvidia's dominance and offering enterprises an open, flexible alternative for deploying generative AI at scale.
The details:
Gaudi 3 promises 50% faster training and inference performance than Nvidia's H100 chip while delivering 40% more power efficiency at a lower cost.
The chips will be available in Q2 of 2024 and are expected to be on par with Nvidia’s H200 line.
Intel also revealed an expanded AI roadmap that includes an open enterprise AI platform, new Xeon processors, and more.
Why it matters: With an already dominant market share and its upgraded Blackwell chip line on the way, it’s doubtful that the Gaudi 3 release has Nvidia sweating. But with AI demand growing exponentially, the chip market is big enough for more than just a single winner.
AI TRAINING
The Rundown: Gamma is a free-to-use AI tool that helps users create beautifully designed presentations in minutes using AI. No design or coding skills are required.
Here is a quick step-by-step:
Sign up for free at gamma.app
On the generated page, select presentation, choose the number of cards you want, and enter a prompt for your presentation.
You’ll receive a structured outline that you can add to or edit according to your needs, and be able to select a theme.
Once you’ve edited it to your preference, you can share it with collaborators, export it, or publish it on the web!
P.S. — Gamma also has a document and website builder that we’ll cover in a future tutorial.
AI RESEARCH
Image source: lmsys
The Rundown: Cohere’s recently launched Command R+ enterprise-grade LLM just ascended to the 6th spot on the LMSYS leaderboard, surpassing competitors like GPT-4 while prioritizing data privacy and security.
The details:
Command R+ is the top open model on the leaderboard, competing with top proprietary models like Claude 3 and GPT-4.
The model performs in line with GPT-4 Turbo and Mistral-Large on multilingual, RAG, and Tool Use benchmarks despite significantly lower costs per token.
Command R+ also matches or outperforms competitors in multilingual evals, offering business capabilities across 10 key languages.
Why it matters: Command R+ is a leap forward for both open and enterprise AI, combining performance with the scalability, data protection, and cost-efficiency that businesses demand. The model’s ascent up the leaderboard is a huge win for open models — and marks yet another GPT-4 level competitor.
NEW TOOLS & JOBS
🗒️ Voicenotes- Notetaker that transcribes and provides info on your thoughts
🔍 Morphic- A fully open-source AI-powered answer engine
🛍️ Shopify Magic- AI-powered e-commerce assistant
🎵 Melodisco- Explore and create your own AI-generated music
💬 RedCom AI- Generate personalized comments on Reddit
🔬 Odaptos- Accelerate user testing with AI
QUICK HITS
Google’s Gemini 1.5 Pro model is now available in 180+ countries via the Gemini API, with the company also introducing new features including native audio understanding, system instructions, JSON mode and more.
Microsoft announced a $2.9B investment in Japan to scale its cloud computing and AI infrastructure in the country, along with plans to provide AI skilling to 3M people and open its first Research Asia lab in Tokyo.
OpenAI released a new update to GPT-4 Turbo, with the ‘majorly improved’ model rolling out across API and in ChatGPT.
eBay launched a new AI-powered "shop the look" feature in its iOS app, providing personalized fashion recommendations and outfit inspirations based on users' shopping history.
Google collaborated with Hugging Face to release CodeGemma, a family of open, code-specialized LLMs with three different model sizes.
Meta published a new study establishing knowledge capacity scaling laws for LLMs, finding that models can store 2 bits of knowledge per parameter.
Archetype AI unveiled Newton, a foundation model that fuses multimodal sensor data and natural language to reason about the physical world.
THAT’S A WRAP
SPONSOR US
Get your product in front of over 550k+ AI enthusiasts
Our newsletter is read by thousands of tech professionals, investors, engineers, managers, and business owners around the world. Get in touch today.
FEEDBACK
How would you rate today's newsletter?Vote below to help us improve the newsletter for you. |
If you have specific feedback or anything interesting you’d like to share, please let us know by replying to this email.
Join the conversation