All along, Microsoft’s new Bing was running GPT-4
Both companies refused to specify which version of GPT was being utilized when Microsoft unveiled the new AI-enabled Bing, which it developed on top of OpenAI’s GPT models. They only disclosed that it was a next-generation model, the same model that underpinned ChatGPT. GPT-4, a major improvement to GPT-3.5, was announced today by OpenAI. It turns out that Bing gpt had been utilizing it all along.
Yusuf Mehdi, corporate vice president and chief marketing officer of Microsoft, stated in the company’s statement today, “We are happy to confirm that the new Bing is running on GPT-4, which we’ve customized for search.” “You’ve already experienced an early version of this powerful model if you’ve used the new Bing preview at any point in the last five weeks.”
It’s not a major surprise as there were many speculations when Microsoft unveiled the new Bing that it ran GPT-4 already. Still, it’s interesting to see that Microsoft trusted the model enough to bet its reputation on it and cover the costs of this more sophisticated model. It’s important to note that Microsoft is putting limits on OpenAI’s model and providing more recent data by combining bing GPT 4 with its own Prometheus model.
Following a more difficult start than the firm probably expected (perhaps due to the new Bing’s susceptibility to hallucinations), Microsoft rapidly refined the new Bing in recent weeks. After imposing several early limits, the company is now easing up again. Microsoft increased the maximum number of turns in a conversation to 15 just yesterday, and it now permits users to have up to 150 chats daily.
Therefore, simply visit Bing to test the new GPT-4 model (or, if you haven’t already, join the waitlist).
By the way, Bing continues to maintain that it is not compatible with GPT-4. Someone needs to inform Sydney that the NDA no longer covers it.