OpenAI releases image-reading AI 'GPT-4 Turbo with Vision' to the public, supporting JSON mode and function calls



OpenAI has begun public release of GPT-4 Turbo with Vision, a multimodal AI capable of reading images.




If you check the model page in the documentation, you will see that the reference to 'gpt-4-turbo' has changed to 'gpt-4-turbo-2024-04-09' at the same time as the introduction of 'gpt-4-turbo-2024-04-09'.



The publicly released GPT-4 Turbo with Vision has been available as a preview version until now. OpenAI's Developer X account has introduced apps created using the preview version of GPT-4 Turbo with Vision.

AI startup Cognition has developed an AI engineer called 'Devin' who can now use GPT-4 Turbo with Vision to perform various coding tasks. A demo movie has been released in which Devin checks the documentation and code by simply asking Devin to 'fix this problem,' and finishes the code while handling the necessary dependencies.




Healthify , an AI health and fitness app developer, used GPT-4 Turbo with Vision to build a service called “Snap,” which provides users with nutritional information through photo recognition of foods from around the world.




tldraw is a service that allows you to draw UIs on an infinitely expanding canvas, and by using GPT-4 Turbo with Vision, you can now automatically code the UI you draw.




According to Owen Campbell-Moore, project manager at OpenAI, the new version not only processes images, but also offers significant improvements in traditional functions such as mathematics.


Related Posts:

in Software, Posted by log1d_ts