ChatGPT
ChatGPT is an AI tool with different variants based on various OpenAI large language AI models such as gpt-3.5-turbo, gpt-8k and gpt-32k. In addition to being a Large Language Model ChatGPT is also a Visual Model that can understand relations and logic inside images. Looking at the progress evolution of OpenAI models we can expect future models to be able to understand videos. In which case, future OpenAI gpt models will be able to train on Youtube videos and understand concepts in videos or in other words motion pictures. This is a great AI tool that can help with quite many natural language tasks.
Features:
-
Webapp: Useful for GUI use cases.
PlaygroundAI: Useful for testing parameters.
History: Convenient chat history provides reference for previously accessed data
System prompts: You can tell ChatGPT to act or response in certain way or style.
Fine Tuning / Tips:
- model: assign model
- temperature: focus of response
- max_tokens: Max. allowed tokens
ChatGPT Pros:
- intelligent
- adapts
- remembers
- affordable
- generous free tier
ChatGPT Cons:
- can be wrong
- no citations
- often has demand related issues
- offline data dating back to 2021
- Chat history doesn't have export or search functions
ChatGPT Price:
App pricing information for ChatGPT is as below:
Testimonials:
Best model I've ever seen.
- User
More Details:
New Update:
Approximately 6 months after its initial release, Chatgpt's underlying model gpt-3.5-turbo has received some serious updates. On the financial side, gpt-3.5-turbo-0613 indicating the launch date June 13, will be 25% cheaper than before. Additionally, very welcome two technical updates are
1- Maximum Prompt input size increase from 4K tokens to 16K token (4x increased capacity)
2- Function calling ability.
While 16K max input allows a lot more complicated and lengthier prompts to be processed by Chatgpt, new function calling abilities will allow more advanced pipelines and less complicated system prompts when creating for example API applications based on Chatgpt responses. Function calling makes it possible to receive programmatic Json-format responses from ChatGPT, a data structure which programmers love.
OpenAI announced that the chages including cost reduction to the default 4K model are effective immediately. gpt-3.5-turbo will automatically point to gpt-3.5-turbo-0613 stable version starting from June 27. And the 16K model can be used by utilizing gpt-3.5-turbo-16k. Check out Holypython's beginner-friendly yet comprehensive Chatgpt development tutorial which can be the first step to your ChatGPT development journey.
