Is GPT-5 Releasing Soon? Sam Spills Beans

The most awaited GPT-5, an AI model is going to be released soon, Sam Altman, CEO of OpenAI confirmed in an interview with Financial Times and said it will possess superintelligence.

That’s very exciting to see some hints about the release of GPT-5. Sam earlier rejected all the rumors if there will be any upgraded version of the current GPT-4 Turbo despite the rumors originated from the person who has a good track record of predicting happenings in the technical world.

GPT-5 Release: What Happened Earlier

Earlier Siqi Chen’s tweet went viral that claimed OpenAI is working on training GPT-5 and that the training will be completed by the end of the year 2023.

And since the training will get completed by 2023 (according to Siqi), the analysts were anticipating GPT-5 may be released in the next few months.

GPT-5

So the year 2024 should have been the release year of 2025.

However, in April 2023 during an event at MIT, Sam confirmed the company is focussing on improving the current GPT models and GPT-5 wouldn’t be available for some time.

GPT-5 Release: What is Happening Now

Now the Financial Times interview with Sam Altman, CEO of OpenAI reveals that the company is working on some model that will possess “Superintelligence”. It can be the next GPT model, ie GPT-5.

Though Sam didn’t specifically use the word “GPT-5”.

So, we shouldn’t jump with the excitement until the clouds of doubts get cleared.

Sam in this interview said, that the company is working on something that will possess superintelligence and that it will require more funds to develop it.

The company will be required to raise more money to fund this model from their existing long-time partner-Microsoft which has already invested more than $10 billion dollar in the company.

Not just the funding, it will require tons of resources as well such as computers, training on trillions of billions of datasets, and extensive research, fine-tuning, and several phases of testing.

Is GPT-5 Releasing anytime Soon?

OpenAI launched ChatGPT-3.5 in November 2022 and after a few months, the upgraded model GPT-4 was announced which was more powerful than the earlier version and trained on larger datasets.

This is when the company started to earn back something to meet with expenses required to keep the AI tool alive.

According to the First Post, the AI bot is costing about $700k per day to the company and the company may go bankrupt soon if it continues.

The GPT-4 model isn’t free and the users have to pay $20/month to continue to use it.

Later a lot of features were added to the AI bot such as Plugins, iOS App followed by the Android App, etc. with the introduction of the latest GPT-4 Turbo.

GPT-4 Turbo is more powerful than GPT-4 and can generate 300 pages of content with a single prompt which is equivalent to a 128k context window.

Not just that GPT-4 Turbo is trained on the data until April 2023.

So its kind of the latest as compared to the first GPT version that was training on data until September 2021.

GPT-4 subscribers can use the Turbo version as they use GPT-4.

Here is the chronology:

  • GPT-3.5 – November 2022
  • GPT-4 – March 2023
  • GPT 4 Turbo – November 2023 (it is not a fresh release though)

The company has been constantly releasing new and upgraded versions of its existing GPT models, and Sam’s recent interview with the Financial Times hints about a new model possessing superintelligence gives us hope for the release of GPT-5 in the near future.

If everything goes fine, such as the company gets funding, secure required resources, we can expect GPT-5 sooner.

Leave a Comment