😺 GPT-4o mini has arrived

PLUS: Our honest take on...DOGS...

Welcome, humans.

Here at The Neuron, we’re all about AI and cats. Sometimes people ask us, what about dogs? Short answer: it depends.

But when we saw this tiny bundle of cuteness overload (AI vid of a mini baby goldendoodle!), we HAD to share it. Can such a potent dose of cuteness be legal?! We finally understand the phrase “cuteness aggression”, because we’re feeling it!

Now, for something not so cute, check out this AI trailer for a Jabba The Hutt movie. (y’know, to “balance the force”.)

Here’s what you need to know about AI today:

OpenAI’s latest model is here: ChatGPT-4o mini!
Meta withdraws models from Brazil, halts new EU releases due to regulations.
Salesforce launched a new customer-service chatbot.
A new startup called Pindrop raised $100M for voice deepfake protection.

ChatGPT-4o mini just launched, and it’s perfect for fast requests.

The newest version of ChatGPT is out. It’s called “ChatGPT 4o mini” and it's apparently the most “capable and cost-efficient model” available today (try it yourself here).

Today’s launch comes at the end of a huge week for small models. Nowadays, everyone from Microsoft, Google, Mistral, Anthropic, to Cohere are all providing their own mini versions.

Why so many small models? To put it simply: smaller models = lower costs. Big companies need smaller models to use and offer AI at scale. Pilot projects and experiments are fine at today’s prices, but actual deployments? Wayyyy too expensive.

GPT-4o mini is 60% cheaper than 3.5 turbo to run (and it’s smarter, too).
Sam Altman said that the best model from 2022 was way worse and “cost 100x more” than GPT-4o mini.

Some speculated that this new launch might be OpenAI’s attempt to bury cheaper competitors like Claude Haiku or Gemini Flash.

Here’s why this is really important: Large models like GPT-4 cost a ton to build ($100M+), and so the people who make them charge more for their best models (that’s why they limit GPT-4o usage on the free plan).

Meanwhile, smaller models are trained on less data, are cheaper to run, and work fine for specific tasks. Some examples:

Mr. Cooper (a mortgage company) and TD Bank are testing midsize models for call centers to analyze voice data and improve efficiency.
WPP (a marketing company) is using Google Flash for tasks like analyzing shopping habits and writing product descriptions.

That said, we’ll still need big models for some more complex tasks, like discovering the formula behind the world’s most gripping mystery novels by analyzing every book written by Agatha Christie.

Our take? Use GPT-4o mini when speed matters more than quality. Here are a few tasks we tried it on:

Coded a Chrome extension to “show me pictures of cats” (<8 seconds)
Wrote a children’s story about a tiny goldendoodle named Goldie (<5 seconds).
Summarized an article and a whole email thread in crazy record time (<4 seconds).
Counted the frequency of “yes”, “no” and “mid” in a spreadsheet dataset (<3 seconds).

However, when we asked it to run the same spreadsheet test on 1000 rows of data, it only counted ~300 total. So avoid GPT-4o mini for complicated tasks involving large datasets.

FROM OUR PARTNERS

Top sales teams are using Attention to win. Why aren’t you?

If you want to close more deals and generate more revenue, then you need to check out Attention’s NEW AI assistants.

Whether you need an automated Sales Manager (Ana), Data Analyst (Dax), or Sales Operator (Tessa), their AI assistants do it all.

Over 100 sales teams have used Attention’s AI tools to do everything: automate their CRM updates, identify actionable insights across calls, and boost overall team performance.

Seriously. One user said “Thanks to Attention, our reps save 20-30 minutes per call with automated CRM data entry.”

If you sell anything, we highly recommend checking Attention out here.

Around the Horn.

All the big tech firms building AI are joining forces to make AI more “secure by design.”
Meta is playing hardball with regulators—it will remove its models from Brazil and won’t offer new models in the EU due to regulatory hurdles.
Miro’s new Intelligent Canvas (waitlist only for now) will use AI to generate diagrams, summaries, feedback, and interactive workflows directly on the whiteboard.

Treats To Try.

RewriteBar offers quick grammar corrections, writing style tweaks, or text translations on any MacOS device. (

*This free guide from Vellum will teach you exactly how to launch production-ready AI products. Get FREE exclusive case studies, prompt engineering advice, and evaluation metrics here!
Proton Scribe is a private writing assistant to write / proofread emails for ProtonMail, which is a security focused Gmail alternative.
Einstein Service Agent is Salesforce’s new customer-service chatbot, building on the Einstein CoPilot (AI for internal teams) it launched earlier this year.
Pindrop provides security to protect your identity from voice deepfakes (raised $100M).
Artificial Agency has an AI-behavior engine for game devs to create NPCs who can improvise in real time, and it can be customized to fit any game project (raised $16M).
Merly builds software to “build better software”, supports 15 programming languages, and works on iOS, Android, Windows, MacOS, and Linux (raised $6.8M).

*This is sponsored content. Advertise in The Neuron here.

Intelligent Insights.

Learn how to combine multiple open-source AI models to get better results than ChatGPT in 3 minutes (link).
AI could get a lot more efficient, not just from smaller models, but from a totally new architecture called test-time-training—very geeky read, but cool idea (link).
AI has essentially “co-opted” the sparkle emoji—and why that matters (link).
Here’s a deep dive on Sam Altman and Arianna Huffington’s new health startup and whether or not AI has become a “technology of faith” (link).
Good thread about how LLMs are impacting the jobs of software devs in the field (link).
An interview with The Atlantic’s CEO on why the company decided to sign a deal with OpenAI, and whatever search project the company is building (link).