• OpenAI partnered with Microsoft to develop GPT’s capabilities.
• GPT 4 was upgraded and refined using the user’s feedback.

Following the Google Workspace AI announcement on Tuesday, OpenAI released the upgraded version of its generative pre-trained transformer system, GPT 4. The current generation GPT3.5, which is deployed in OpenAI’s popular ChatGPT, can only read and reply to the conversation. While the new upgraded version of GPT 4 can generate text for image input. 

The OpenAI team stated:

“GPT 4 will be more creative, reliable, and able to handle much more nuanced instruction than GPT 3.5.”

OpenAI has partnered with Microsoft to improve GPT’s capabilities. According to the report, GPT has spent the last six months upgrading and refining the system. Moreover, the company reported that GPT 4 passed the simulator test with a score around the top 10, compared to GPT 3.5 scored in the bottom 10. 

GPT 4 will be available on both ChatGPT and API. The users have to subscribe to ChatGPT plus to get GPT 4 access. API access for the GPT 4 will be handled through the waitlist. 

Upgraded Features of GPT 4

In GPT 4, users can add visual inputs in the chat. It accepts both text and images as input to generate output. ChatGPT will now summarise the various images or detail into words with the best understanding. It can iterate with users on creative writing tasks and can help to compose music. This new version can handle 25,000 words. GPT 4 has human-level performances. OpenAI incorporates more human feedback into the GPT 4. This includes the feedback submitted by the users to improve the ChatGPT.  

GPT 4 is already being used by popular firms. To highlight, the payment platform Stripe uses GPT 4 as virtual assistance and to combat fraud. Duolingo for a more immersive and flexible conversation experience. Morgan Stanley to organize its vast knowledge base.

Recommended For You