OpenAI debuts GPT-4 Turbo and enhanced fine-tuning program for GPT-4
Copyright shield of OpenAI will defend customer from copyright infringement
 |
Image: Franz Bachinger via Pixabay |
Earlier today, OpenAI at its first-ever developing conference, introduced GPT-4 Turbo, providing advanced and improving AI-generating text features. It is currently available for all paying developers to try gpt-4-1106-preview in the API and a stable production-ready model will be available in the coming weeks.
GPT-4 Turbo is cheaper to run for developers
On March 2023, AI platform OpenAI released GPT-4( accepting image and text inputs), the company introduced the GPT-4 turbo version an improved and less expensive version of its predecessor flagship AI model. They're pricing at input tokens一 the basic unit of text or code that an LLM AI uses. The turbo version costs $0.01 per 1,000 input tokens 一 3x cheaper than its flagship version GPT-4 costs $0.01, and 2x cheaper for output tokens (generated by a model based on input tokens).
The upgraded version is still able to take image prompts
as input. OpenAI introduced the DALL·E 3 model for image generation. The
pricing of the GPT-4 turbo model also depends on the input image size. For
instance, passing an image with 1080×1080 pixels to GPT-4 Turbo costs $0.00765,
as the blog post reads. OpenAI also introduces Test to test-to-speech model for the
newest AI model.
The GPT-4 Turbo model takes edge on the GPT-4 with the
most recent knowledge 一 Turbo's version
comes with knowledge up to April 2023. That means it should provide more
accurate recent information (cut off to limit) than GPT-4 一 as knowledge up to the web search of September 2021.
Users now will be able to request more questions than
GPT-4, but the price remains the same. To help you scale your applications,
we’re doubling the tokens per minute limit for all our paying GPT-4 customers,
Open AI says in the blog post.
GPT-4 Upgrades
GPT-4 Turbo features a 1,28,000 context window which is the largest window ever for an AI model available in the market. So customers will able to see more data. OpenAI claims that the newest AI model can fit more than 300 pages of text in a single prompt. Previously, the GPT-4 model came with a context window of 8,000 and 32,000 which is four times smaller than the turbo's model.
The company OpenAI claims that GPT-4 Turbo is capable of
performing complex tasks in one prompt. With user-specified programming
language, code can be implemented most efficiently. such as JSON, XML, C++, and
Python. Code Interpreter writes and runs Python code in a sandboxed execution
environment, and can generate graphs and charts, and process files with diverse
data and formatting, OpenAI writes.
Copyright Shield
OpenAI not only provides improvements in GPT-4 but also provides safety from copyright infringements named Copyright Shield. The company will pay customers in case of legal action face due to copyright. But as announced this is just for the ChatGPT Enterprise and to developers using ChatGPT's application programming interface. Users of the free version will not included under Copyright Shield.