OpenAI is pushing new features in its latest beta version for ChatGPT Plus members, as per a report by The Verge on Monday. The features include the ability to upload and analyse files to work with them, and multimodal support. This means that the chatbot will now guess what users want based on the context instead of having to select modes like Browse with Bing from the GPT-4 dropdown.
The features are similar to the ChatGPT Enterprise version that OpenAI released in August. The report showed that while the user was able to test out the Advanced Data Analysis feature, they still didn’t have access to the multimodal features.
Once the file is uploaded, it takes ChatGPT few moments before it starts working with it and then produces a summary of the data in it. It can also answer questions and generate data visualisations based on prompts.
The function isn’t just for text files. Subscribers posted screenshots of a conversation where they uploaded an image of a dog or a capybara and asked ChatGPT to create a different variation of the image, like a Pixar-style version, using OpenAI’s image generating model DALL-E 3.
(For top technology news of the day, subscribe to our tech newsletter Today’s Cache)
OpenAI had introduced image and voice capabilities with ChatGPT earlier this month.
The Microsoft-backed AI firm had also been building a new AI model called Arrakis which was expected to lower the costs of running the compute-intensive GPT-4, a report by The Information outlet stated. A couple of weeks ago, however, it was reported that OpenAI had shelved the project due to quality constraints. Arrakis was set up to be a challenger to Google’s yet-to-be-released multimodal AI model Gemini.
