(CTN News) – OpenAI makes minor adjustments to its models and pricing periodically, and today is one such occasion. Several new models have been released, and API access has been reduced – this is primarily of interest to developers, but also serves as a bellwether for future consumer products.
GPT-3.5 Turbo is the model most people interact with, typically through ChatGPT, and it serves as ChatGPT’s OpenAI, why bother?standard – if your answers aren’t as good as
It is also a popular OpenAI API because it is cheaper and faster than GPT-4 in most cases. As a result, paying users will be pleased to learn that input prices have fallen by 50% and output prices by 25% to $0.0005 per thousand tokens in, and $0.0015 per thousand tokens out.
When people use these APIs for text-intensive applications, such as analyzing entire papers or books, those tokens begin to accumulate.
When open source or self-managed models catch up to OpenAI’s performance, the company needs to ensure that its customers do not simply leave. Therefore, the prices have steadily declined – though this is also a natural consequence of the streamlining of models and the improvement of their infrastructure.
The GPT-3.5 Turbo has also received a new version, 0125 (i.e. today’s date), that includes “various improvements” but which OpenAI thought were not worth mentioning. It is surprising that they do not mention more information about the last version, 0613.
There is a new preview version of GPT-4 Turbo for API use, also 0125, containing the following interesting fix:It is possible that the model absorbed and embodied the wrong lessons from the developers whose habits it ingested. There is a possibility that AI is already quitting quietly. AI is a good thing!
It is still in preview mode. However, GPT-4 OpenAI Turbo with vision (i.e., GPT-4 V) will be available to the general public within the next few months.
In addition, there are a few new and improved text embedding models for those who are more technical. Additionally, the company has released a new version of its free moderation API that identifies potentially harmful text. When you are experimenting with using this to augment your moderation needs, look for version 007.