• Source:JND
Add as a preferred

OpenAI, a popular AI research and deployment firm, has announced that GPT-4 Turbo with Vision is now rolling out in ChatGPT, the company's viral AI chatbot. The capabilities are also available in the Application Programming Interface (API). Via the feature, paid accounts in ChatGPT can input an image to generate responses. Here are more details about GPT-4 Turbo with Vision.

GPT 4 Turbo with Vision

GPT-4 Turbo, the company's latest LLM, supports up to 1,28,000 tokens which is 8x the tokens of GPT-3.5 Turbo. For context, tokens are the chunks of data fed into the model as input. The newest model is up to date with information up to December last year, offering more relevant and updated information.

Also Read: GPT-4 Turbo vs GPT-4: What's The Catch In New 'More Capable' Yet 'Cheaper' Next-Gen AI Model; Comparison 

"GPT-4 Turbo with Vision allows the model to take in images and answer questions about them," OpenAI said in its blog post. It can also accept media input via a Uniform Resource Locator (URL). While the model is the company's most advanced piece of tech, it does have a range of limitations.

As per OpenAI, it cannot correctly process medical images (such as CT scans), images with non-English or small texts, images with graphs and stylised texts, panoramic and fisheye images, and more. The company has blocked the usage of the system for submitting CAPTCHAs. Users can upload "PNG (.png), JPEG (.jpeg and .jpg), WEBP (.webp), and non-animated GIF (.gif) file formats and the size is restricted to 20MB/image.

Also Read: Copilot AI: GPT-4 Turbo To Code Interpreter, Microsoft Announces New Features As AI Engine Set To Turn 1; Know Here

"We will delete the image for you automatically after it has been processed by the model," OpenAI added. Other capabilities of GPT-4 Turbo include making websites from scratch, coding and more. GPT-4 Turbo is available to ChatGPT Plus users, the price of which starts from $20/month (about Rs 1,660) in India.