OpenAI launched GPT-4o and a MacOS app

PLUS: $16K humanoid robot

In partnership with

Good morning, human brains, and welcome back to your daily munch of AI news.

Here’s what’s on the menu today:

  • GPT-4o. Woah or no? 🤖 🔥

    OpenAI’s model understands text, images, audio, and video.

  • A foldable, portable, $16K humanoid robot 🦾 🦿

    The G1 runs, completes chores, prepares food, and more.

Sponsor Bot Eat Brain (I discounted it) 🤯

MAIN COURSE

OpenAI’s ridiculously hyped event 🤖 🪞

Yesterday, OpenAI launched GPT-4o. It’s a new multimodal AI model that processes audio, vision, text, and video in real-time.

What can I do with it?

You can now have voice conversations with GPT-4o directly from your computer using Voice Mode. You can ask it questions about something on your screen, video from your camera, and more.

It’s better than GPT-4?

Yes. It set new records in general knowledge, speech recognition, translation, and visual perception. It matches GPT-4 in English text/coding tasks and significantly improves multilingual performance at half the cost.

How much does it cost?

It’s free for everyone to use. Paid users get 5 times more usage capacity.

Anything else?

You bet. OpenAI is rolling out a ChatGPT desktop app for Mac users. If you have Windows, you’ll get one “later this year.”

Have an AI Idea and need help building it?

When you know AI should be part of your business but aren’t sure how to implement your concept, talk to AE Studio.

Elite software creators collaborate with you to turn any AI/ML idea into a reality–from NLP and custom chatbots to automated reports and beyond.

AE Studio has worked with early stage startups and Fortune 500 companies, and we’re ready to partner with your team. Computer vision, blockchain, e2e product development, you name it, we want to hear about it.

SIDE SALAD

Folding out for a hero… 🦾 🦿

On Sunday, Unitree Robotics unveiled the G1. It’s a $16,000 humanoid robot.

I gave my crush an origami flower. She said it was FOLD-fashioned... 🤦‍♂️

What does it do?

In the demo, the G1 solders wires, cooks food, twirls batons, and more. It handles delicate objects, recovers from punches/kicks, and completes household chores with impressive dexterity and precision.

What’s special about it?

Its “hollow joint” wiring and several motors allow it to move, bend, and fold into a portable package. It can run, open bottles, use tools, and more. It leverages imitation and reinforcement learning, which makes it highly adaptable and useful.

How big is this thing?

It’s 50” tall and weighs 103 pounds. For a humanoid robot, it’s compact and easy to transport.

Why do I care?

The G1 is expected to accelerate humanoid robot development and pressure other companies to lower their prices. It has game-changing potential for household and industry applications.

I tried blind-FOLD-ed archery... I didn’t know what I was missing… 🤪

WHAT WOULD YOU DO?

“Mom, can we keep them?” 👧

“You can only have one!” 👩

Pick your new pet. 👇

Choose your mythical creature.

And no, you can't have them all.

Login or Subscribe to participate in polls.

Hope you’re on my side. 😳

Ideas? Comments? Complaints?

or Respond to this email 📧

Now, onto the daily munch. 🤗

YOUR DAILY MUNCH

Think Pieces 🧠

Zuckerberg’s warning? He claims that the GPU shortage is easing, but energy constraints will soon limit AI growth.

An indestructible robot hand? Google DeepMind is testing a highly durable robot hand designed to withstand extreme damage during the AI training process.

Startup News 💰

Neuralink's brain-computer interface experienced mechanical issues. It showed a decline in its data transfer rate due to its thread design.

Nvidia announced its open-source CUDA-Q platform. The goal is to advance quantum computing research and integrate QPUs (quantum chips) in its centers.

Research 👨‍🔬

BASS — improves the speed and efficiency of generating multiple sequences in AI language models by optimizing speculative sampling and GPU utilization.

PuLID — a tuning-free method for customizing identities in text-to-image generation, ensuring high fidelity and minimal disruption to the original model.

MEMES FOR DESSERT

TWEET OF THE DAY

Early GPT-4o users report it’s less “lazy.” There’s a lot of hype around this model, so it’s too early to tell for sure.

Tag us on Twitter @BotEatBrain for a chance to be featured here tomorrow.

AI ART SHOW

Until next time 🤖😋🧠 

What'd you think of today's newsletter?

Login or Subscribe to participate in polls.