Advertise with us

Welcome, AI enthusiasts.

All eyes are on Nvidia’s earnings report today, but relative newcomer Cerebras has an announcement that might just beat the world’s second-largest company’s numbers. And: We finally get a sneak peek into OpenAI’s “Strawberry.”

In today’s AI rundown:

• Cerebras releases the world’s fastest AI inference system

• Tutorial: How to create AI images with Meta AI

• From the Frontier: ‘Strawberry’ details revealed

• Everything else you should know today

• 3 new AI tools to boost your productivity

• AI-Generated Images: Old Books

Read time: 4 minutes

NEXT IN AI

Cerebras releases the world’s fastest AI inference system

We’re still in the AI dial-up era, but a startup called Cerebras wants to do to LLMs what high-speed internet did for web browsing. Earlier this year, it showed off the world’s largest AI chip (it’s about the size of a dinner plate). Now, it’s releasing a new system that can run AI products via the cloud — and at unprecedented speeds.

How it works: Cerebras packed its record-setting chips onto a system called CS-3, then used that infrastructure to build some of the world’s largest supercomputers. Its latest release helps companies put those LLMs to use in the real world. 

What’s inference anyway? 

• It’s the process of taking in new information, then running it against a dataset that the model was previously trained on

• It can be used to spot patterns in large swathes of data, and it can help models make decisions much faster than other approaches

• Inference already makes up about 40% of the AI hardware market, but that figure is steadily ticking up

Why is Cerebras so much faster than its rivals? Traditional GPUs have to interact with external memory each time they crunch a piece of data; but because Cerebras’ chips are so massive, there’s room to fit a ton of memory directly onto them, completely bypassing that step.

The results: Many performance-focused systems have to scale back their accuracy in order to boost speed. But Cerebras says its architecture runs at a native 16-bits, meaning its precision never drops off. When it comes to training Meta’s Llama 3.1, it’s around 20 times faster than comparable Nvidia GPU-based systems — at just one-fifth of the cost.

PRESENTED BY ASSEMBLY AI

AI Agents now join your Zoom and run your meetings

Spinach AI runs daily standups and project meetings for thousands of companies.

• Focused meetings - runs the meetings, keeps track of time

• Accurate summaries - saved in Google Docs, Notion or Confluence

• Ask questions - “what are the open action items from last week?”

• Speaks 100 languages

Spinach offers a 14-day trial and takes 30 seconds to set up**.

Get Started Now**

THE AI ACADEMY

Create AI images on WhatsApp with Meta AI

Open WhatsApp and click on the Meta logo at the top of the screen.

• It will open a new chat window for you.

• Explain what you want to generate and watch the magic happen.

• It will generate images for you in real-time.

• You can share your creations with your friends and family on WhatsApp and enjoy.

Prompt used: Imagine a cute golden retriever in front view in a park with his 40-year-old owner, a lady, and children playing with him. It's golden hour, and beautiful sun rays are striking from the background.

FROM THE FRONTIER

Is it Strawberry Season?

Details about OpenAI’s “Strawberry” –  a rumored model that could take AI reasoning to the next level – have finally been revealed.

Here they are:

• Sources told The Information OpenAI might integrate Strawberry into ChatGPT, instead of releasing it as a standalone model

• It’s reportedly so powerful that a team showed it off to American national security officials this summer

• It would be able to perform high-level math and logic problems — even those it was never trained on, although it’d take longer to generate results

• It could be released as soon as this fall, with a new model code-named Orion coming at a later date

• The company is struggling to raise more capital, so this could be just the boost it needs to power through

PRESENTED BY GUIDDE

Create video documentation 11x faster with AI

Tired of explaining the same thing over and over again to your colleagues? Guidde is a GPT-powered tool with AI-generated documentation that helps you explain the most complex tasks in seconds.

The best part? Our extension is free. Try it here

Trending AI Tools

• ✅ Kerlig: An AI-powered writing assistant that can be used in Slack, Figma, Gmail, LinkedIn, and more.

  ✅ Arold: Use AI to reply to guests within your Airbnb inbox from a single tap.

  ✅ PackPack: An AI-driven bookmark management tool tailored for saving content from online resources like news and social media.

_{* indicates a promoted tool, if any}

AI around the world

• Answers in a Flash: Google is rolling out three new Gemini variants, including a more powerful Pro model that can tackle complex coding and logic problems.

• Unexpected Blessing: Elon Musk has endorsed California’s controversial AI safety bill, arguing governments should monitor LLMs “just as we regulate any product/technology that is a potential risk to the public.”

• Appliance Upgrade: Samsung’s touchscreen refrigerators are getting new AI capabilities, while its AI TVs will now receive seven years of updates.

Join the AI Revolution

Fetch AI News is a premier AI Newsletter with over 50000 AI enthusiasts globally, including professionals from top-tier companies like OpenAI, Google, Meta, and Microsoft.

What We Can Offer:

• Introduce new products or features

• Launch an impactful advertising campaign

• Conduct targeted surveys to gather valuable insights

• Any other business cooperation opportunities

Thank you for being part of the AI Insights Today community! Help us spread the word about the latest in AI by sharing this newsletter with your friends and family. Together, let's uncover the future of technology.