OpenAI o1 Is Out: The Strawberry AI That Thinks Like a Human Aug/1/26

@mihwa.jaeseung

11.9K followers · usually 2.8K views a post

Views

21.3K

Likes

278

Comments

15

Shares

46

SoundTikTok sound · original sound - mihwa.jaeseung

Promoted product

OpenAI o1 (Strawberry)

Why it worked

The video likely performed well due to its timely announcement of a significant AI development from a major player (OpenAI) and its focus on a novel feature (reasoning tokens) that sparks curiosity. The "native_mention" approach, where the product is the subject of the content rather than a hard sell, also contributes to organic reach.

Summary

The video announces the release of OpenAI's new AI model, "Strawberry AI" or "o1". The speaker explains that this model is fundamentally different from previous LLMs, incorporating a time-based component and reasoning tokens to improve response generation. The speaker suggests that this design choice, while not fully disclosed due to its closed-source nature, results in a smarter model capable of handling complex tasks more efficiently.

Structure

1Introduction of OpenAI's new AI model, "Strawberry AI" (o1)
2Explanation of its fundamental difference from previous LLMs
3Highlighting the addition of a time-based component and reasoning tokens
4

Original caption

OpenAI o1 Is Out: The Strawberry AI That Thinks Like a Human #OpenAI #o1 #StrawberryAI #ReasoningRevolution #TechNews

More from @mihwa.jaeseung

Product placement

The video discusses and shows visuals related to OpenAI's new AI model, "Strawberry AI" (o1), highlighting its features and capabilities.

On-screen text

OpenAI STRAWBERRY YouTube OpenAI o1 Is Out: The Strawberry AI That Thinks Like a Human Aug/1/26 PREVIEW 01 STRAWBERRY CURSOR REPLIT APP CODE ENTIRE APPS! #adaptAndFly A Scary Smart new AI Model OpenAI's o1 (Strawberry) INSURANCE & Financial Professionals can't ignore 7 MIN READ Panos Leledakis - IPAN

Transcript

OpenAI's strawberry has dropped. They call it 01 because of course they did. Why pick a name that's easy for people to remember? And I want to give you a quick preview on what's in the box. So, at the end of the day, this is a fundamentally different model and that's the first thing to understand. This is not a bigger LLM. This is a different kind of model that has an LLM component to it. And so what they've done is fundamentally change the architecture of how the model responds. They we have added a time-based component that allows the model to use reasoning tokens that it does not show to you to compute a better response. Now, that's a design choice. They didn't have to hide the reasoning tokens, but I think they decided to because it's a closed source model and they don't want to reveal too much of the secret sauce. At the end of the day, what this feels like is it feels like a fundamentally smarter model. I can give it things that would be difficult for me to to do in hours at all. And so, what this feels like is it feels like a fundamentally smarter model. I can give it things that would be difficult for me to to do in hours at all.