GPT-4.5 is not the AGI moment OpenAI teased

Getting your Trinity Audio player ready...

GPT-4.5 is almost here. Sources within OpenAI and Microsoft (NASDAQ: MSFT) have revealed that the latest model of the GPT family of AI models could launch as soon as this week, but despite founder Sam Altman’s artificial general intelligence (AGI) claims, it won’t be the massive leap he has touted.

OpenAI has been working on GPT-4.5 and GPT-5 since last year and was originally set to launch the former in December. Codenamed Orion, GPT-4.5 has been teased by some executives at the AI firm as being 100x more powerful than GPT-4 while using similar computation resources.

OpenAI was taking its time to announce Orion’s release date, until Elon Musk shook things up by launching Grok 3, xAI’s most powerful model, which the Tesla (NASDAQ: TSLA) CEO says is “the smartest AI on earth.”

In a quick response, Altman claimed that Orion “has been much more of a ‘feel the AGI’ moment among high-taste testers than I expected!”

AGI is an advanced form of AI that can learn and apply knowledge across a broad range of tasks in a human-like manner. Unlike existing AI applications designed for specific niches, such as conversational chatbots like ChatGPT, AGI can reason, solve complex challenges, and adapt without new programming.

Not yet AGI

Despite Altman’s proclamation, OpenAI hasn’t cracked AGI yet and is unlikely to with GPT-4.5 or GPT-5, an improved version expected in May.

In fact, Altman has been adamant that AGI is not as close to reality as some optimistic AI developers have posited. In a January 20 post, he wrote, “We’re not going to deploy AGI next month, nor have we built it.” He urged ChatGPT users to cut their expectations 100x.

twitter hype is out of control again.

we are not gonna deploy AGI next month, nor have we built it.

we have some very cool stuff for you but pls chill and cut your expectations 100x!

— Sam Altman (@sama) January 20, 2025

While it may not achieve human-level reasoning capability, GPT-4.5 is still a major improvement for OpenAI. Altman says it will be the company’s last “non-chain-of-thought” model. A chain-of-thought AI model explicitly breaks down its reasoning process, unlike existing GPT models, which only provide the final responses.

Altman’s ultimate goal is AGI, and while it may not come with GPT-4.5, it’s edging closer. In a January blog post, he claimed that OpenAI is “now confident we know how to build AGI as we have traditionally understood it.”

“We are beginning to turn our aim beyond that, to superintelligence in the true sense of the word,” he claimed.

The company’s greatest leap was the release of the o3 reasoning model last December, which scored 87.5% on the ARC-AGI Benchmark, the highest of any model at the time. This benchmark assesses AI models on their ability to perform human-like reasoning. Humans, on average, score 80%.

For context, when the benchmark launched in 2020, all models were at zero, and over four years later, in mid-2024, they had only climbed to 5%.

GPT-5 is expected to launch in May, with sources telling The Verge that OpenAI wants to align the launch with Microsoft’s Build conference. It will unify all the company’s series’ of reasoning models, such as the record-shattering o3, with the GPT models into one powerful system.

In order for artificial intelligence (AI) to work right within the law and thrive in the face of growing challenges, it needs to integrate an enterprise blockchain system that ensures data input quality and ownership—allowing it to keep data safe while also guaranteeing the immutability of data. Check out CoinGeek’s coverage on this emerging tech to learn more why Enterprise blockchain will be the backbone of AI.

Watch: Transformative AI applications are coming