OpenAI announced new product upgrades to ChatGPT at its OpenAI DevDay event on Nov. 6.

First announced was GPT-4 Turbo, its latest AI model. The new model provides answers with context up to April 2023 instead of January 2022.

OpenAI CEO Sam Altman said:

“We are just as annoyed as all of you, probably more, that GPT’s knowledge about the world ended in 2021.”

GPT-4 Turbo also accepts considerably more input, including inputs that are 300 pages long instead of just 3,000 words long. It also accepts DALLE-3 AI-generated images and supports text-to-speech with six preset voices. It also reduces the cost of input and output tokens to $0.01 and $0.03 (a reduction of 3x and 2x, respectively).

The improved version of GPT-4 Turbo is currently available as a developer preview but will be released to general users in the coming weeks.

Microsoft CEO Satya Nadella also appeared during the event, stating that his firm will ensure that OpenAI has the most computing and best training and inference systems so that it can “aggressively push forward” on its AI development roadmap. Microsoft has invested $13 billion in OpenAI to date and has otherwise supported the firm.

Users can now create custom ChatGPTs

OpenAI also announced that it will allow users to create custom ChatGPT chatbots, as indicated in an announcement from the company itself.

That announcement indicates that users will be able to create individual “GPTs” for a specific purpose without coding. These GPTs can also be shared publicly through an upcoming feature called the GPT Store, similar to other app stores.

That announcement additionally says that OpenAI has set up new privacy and safety measures. It also says that new features for developers and enterprises are available and describes plans to democratize the development of artificial intelligence.

The news comes weeks after reports that OpenAI could potentially reach a valuation of $86 billion and news that the company is introducing a risk preparedness team.