Tools

Search

Explore

Videos Channels Figures

Atmrix

About

Tools

Search

Explore

Videos Channels Figures

Atmrix

About

O

OpenAI Cast

04/11/25

@ OpenAI

Creating a giant model like GPT-4.5 requires a lot of people, time, and computational resources.

Video

O

Pre-Training GPT-4.5

@ OpenAI

04/11/25

Related Takeaways

O

OpenAI Cast

04/11/25

@ OpenAI

The sheer volume of work required for GPT-4.5 was a byproduct of the model's specifications, necessitating changes that couldn't be achieved with the same stack used for GPT-4.

O

OpenAI Cast

04/11/25

@ OpenAI

The initial goal for GPT-4.5 was to create a model that is ten times smarter than GPT-4, which we ultimately achieved through a complex process.

LF

Lex Fridman Cast

02/03/25

@ Lex Fridman

The largest AI models, like GPT-4, require massive GPU clusters, with power consumption reaching unprecedented levels, highlighting the increasing scale of AI infrastructure.

YC

Y Combinator Cast

03/11/25

@ Y Combinator

GPT-4.5 is potentially more than 10 times the size of GPT-4, representing a step forward in both pre-training and post-training processes.

YC

Y Combinator Cast

03/11/25

@ Y Combinator

GPT-4.5 is OpenAI's largest and most humanlike model to date, representing a significant step in scaling up unsupervised learning and achieving a deeper understanding of the world and human experience.

YC

Y Combinator Cast

03/11/25

@ Y Combinator

Despite its advancements, GPT-4.5 is significantly more expensive than previous models, costing 30 times more per input token and 15 times more per output token.

YC

Y Combinator Cast

01/24/25

@ Y Combinator

With the release of GPT-3, it seems the sky is the limit for this new paradigm of scaling large language models.

YC

Y Combinator Cast

03/11/25

@ Y Combinator

The high costs associated with GPT-4.5 make it unsuitable for many users at this time.

O

OpenAI Cast

04/11/25

@ OpenAI

The collaboration between teams on model specifications for GPT-4.5 involved deep co-design efforts, focusing on optimizing functionality and ensuring the ML and systems work well together at scale.