OpenAI has released GPT-4, its latest and most powerful deep-learning model capable of understanding both images and text. GPT-4 is already being used by Stripe to scan business websites, by Duolingo for a new language-learning subscription tier, by Morgan Stanley to retrieve information from company documents and by Khan Academy for automated tutoring. GPT-4s image understanding capability is currently being tested by Be My Eyes Virtual Volunteer feature, which answers questions about images sent to it. The AI model passes various professional and academic benchmarks and is available to OpenAIs paying users via ChatGPT Plus, with pricing of $0.03 per 1,000 prompt tokens and $0.06 per 1,000 completion tokens.
