Background
Member Login
menu

With Less Confidence, OpenAI Introduces 'GPT-4.5' That Is More Powerful And Thoughtful

OpenAI GPT-4.5

The race is on, and it's a tough one.

Since OpenAI introduced ChatGPT, pretty much all tech companies, large and small, realize the potential of this Large Language Models technology. What happened next, is a disruption in various industries, and an arms race that sees a lot of companies becoming rivals.

With many big players in the field, OpenAI is also competing with relatively smaller players, which are apparently more than just a handful.

What this means, the only way to thrive is by improving.

After upgrading its AI to GPT-4, this time, OpenAI introduces GPT-4.5, an upgraded version of the AI model that powers ChatGPT, to select users.

In the announcement, OpenAI said that:

"We’re releasing a research preview of GPT‑4.5—our largest and best model for chat yet. GPT‑4.5 is a step forward in scaling up pre-training and post-training. By scaling unsupervised learning, GPT‑4.5 improves its ability to recognize patterns, draw connections, and generate creative insights without reasoning."

Early testing shows that interacting with GPT‑4.5 feels more natural.

OpenAI said early tests suggest that interacting with this model "feels more natural."

Thanks to its broader knowledge base, improvements in its ability to follow user intent, and its greater emotional quotient should make the model more useful for tasks like improving writing, programming, and solving practical problems.

OpenAI also expects the model to hallucinate less.

Altman described it as "the first model that feels like talking to a thoughtful person."

"I have had several moments where I've sat back in my chair and been astonished at getting actually good advice from an AI," he said.

While CEO Sam Altman described in a post on X that the model as having "a magic to it i haven't felt before," the model isn't actually a huge jump over its predecessor.

Whereas GPT-4o marked a huge improvement over the GPT-3.5 that powered the original ChatGPT, this GPT-4.5 is more like a tweak.

Despite the model requires humongous computing power to operate, with Altman saying that GPT-4.5 is a "giant, expensive model," this upgrade isn't too significant in terms of results.

This is why the model isn't making that huge gains in benchmarks.

Altman said that OpenAI is trying to push the limits, but the company the resources the product needs is escalating, and that it's running out of GPUs.

" [...] we will add tens of thousands of GPUs next week and roll it out to the plus tier then," he said.

Regardless, the release represents progress for one of the most highly prized startups in the world.

The release of the model also comes about a month after DeepSeek-R1 wrecked havoc in the U.S;, plummeting the stock price of GPU-maker Nvidia to nearly $600 billion in a single day.

GPT-4,5 is initially introduced as a research preview to users who pay $200 a month for ChatGPT Pro and developers in the API to "better understand its strengths and limitations."

OpenAI added that GPT-4.5 would be the company's "last non-chain-of-thought model."

It's worth noting that GPT-4.5 may be the most powerful model the company has, but it's totally different to the o1 and the o3 reasoning models

GPT-4.5 is more like a swiss-knife AI that focuses on "word knowledge, intuition, and reducing hallucinations," whereas reasoning models are designed to think before responding, and are better suited for quantitative tasks.

Read: DeepSeek, The Chinese AI That Makes Silicon Valley Nervous And The U.S. Concerned

Published: 
01/03/2025