For example, the company noted in a statement that GPT-4 scored in the top 10% of test takers in a simulated bar exam. However, the improvements touted in the release show an exponential leap forward from GPT-3.5 and GPT-4. It is still flawed, still limited, and it still seems more impressive on first use than it does after you spend more time with it. it is available today in our API (with a waitlist) and in ChatGPT+. Here is GPT-4, our most capable and aligned model yet. In a tweet coinciding with the release, Altman called the latest release 'still flawed, still limited.'

He urged people to use 'great care' when using language model outputs, particularly in 'high-stakes contexts.' As recently as last December, OpenAI CEO Sam Altman admitted that ChatGPT creates 'a misleading impression of greatness' and that it 'hallucinates' facts.

In various statements, the company has reiterated that while GPT-4 exhibits human-level performance in a number of professional and academic benchmarks, it is less capable than humans in many real-world scenarios. However, despite the flurry of headlines and reactions following the release, the company has continued to downplay expectations. The text-input feature is available to ChatGPT Plus subscribers - the technology's first paid plan - and to software developers via a waitlist, while the image-input ability remains in preview. In a statement this week, San Francisco-based OpenAI announced the release of the new, multimodal version of the technology which can generate content based on images as well as text.

Microsoft suites