GPT-5 Coming Soon ??

First up, GPT-4.5 (aka Orion) is on its way and will be the last non-chain-of-thought model before OpenAI unifies its AI lineup. The plan? Merge the o-series and GPT-series so the system can seamlessly integrate the tools and optimise thinking time.

Then there’s GPT-5, which will power both ChatGPT and the API. It will include OpenAI’s latest tech (like o3) but don’t expect o3 to be a standalone release. Altman has hinted that GPT-4.5 or GPT-5 could drop in the next few weeks or months, so let’s wait and watch.

ChatGPT’s free tier will receive unlimited GPT-5 at a standard intelligence level (as long as it is not abused). Meanwhile, the Plus subscribers will get a smarter version, and the Pro subscribers will have access to an even more advanced one. Expect features like voice, canvas, search, and deep research.

It’s Raining Expectations

To give a glimpse of what OpenAI is doing with its latest models, o3 won gold at the International Olympiad in Informatics 2024. According to its research, more training and test-time computation improve model performance, nearing human levels.

Check out this video on how models from OpenAI, Anthropic, and others are changing things for developers.

Meanwhile, to reduce its dependence on NVIDIA, OpenAI is pushing forward with its own AI chips. The company plans to finalise its first in-house chip design soon and send it to TSMC for fabrication.

Also, OpenAI is reportedly in talks with SoftBank for a $25 billion investment, but that’s not even the wildest part. Elon Musk and a group of investors (including xAI, Vy Capital, and Ari Emanuel) made a $97.4 billion bid to buy OpenAI’s nonprofit arm.

Altman’s response? A hard NOPE. He made it clear: "The OpenAI mission is not for sale."?

Despite this, Altman said that the company would like to work with China. When asked whether the US would let him do that, he said, “I know that for sure, no. Should we try as hard as we can? Absolutely, yes.”

Concurrently, Apple has gone ahead and partnered with China’s Alibaba for AI features, while Hugging Face is releasing a new small language model to compete with the open-source giants. Anthropic, too, plans to release its reasoning model in a few weeks, which would compete with OpenAI with its “deep reasoning” and quick responses.

Why Would Musk Stop Then?

Well, given Musk's hype surrounding the upcoming Grok 3 release, it looks like there’s some drama brewing at xAI. He called it “scary smart” and claimed that its reasoning abilities surpass those of any AI model to date. Speaking at Dubai's World Governments Summit, Musk said that Grok-3 would be released in “a week or two".

While Musk is setting high expectations, Benjamin De Kraker, a former xAI engineer, just announced his resignation—not because of the model itself, but because of what he describes as a “bizarre internal censorship” issue.

According to De Kraker, xAI forced him to choose between deleting a harmless post mentioning “Grok 3 (TBD)” or getting fired. The kicker? Grok 3 has already been officially acknowledged by xAI, Musk, and an xAI blog post. Yet, the company claims that simply referencing it is a breach of confidentiality.?

Though De Kraker’s firing isn’t confirmed yet, he refused to delete his post, arguing that it was just a personal opinion and contained no secret information. He also hinted that xAI might have been frustrated that his post didn’t automatically rank Grok 3 as the top AI model.

De Kraker still hopes xAI and Musk succeed, but he’s disappointed that a company that touts free speech would try to silence an employee over a trivial opinion.