OpenAI is always making small adjustments to its models and pricing, and today is no exception. The company has released a handful of new models and reduced the price of API access — this is primarily of interest to developers, but also serves as a bellwether for future consumer choices.
GPT-3.5 Turbo is the model that most people interact with, usually through ChatGPT, and acts as a sort of industry standard now — if your responses aren’t as good as ChatGPT’s, why bother? It is also a popular API, lower cost and faster than GPT-4 in many tasks. So paying users will be happy to hear that entry prices are being reduced by 50% and output by 25%, to $0.0005 per thousand tokens and $0.0015 per thousand tokens.
As people play around with using these APIs for text-intensive applications like parsing entire documents or books, these tokens really start to add up. And as open-source or self-managed models catch up to OpenAI’s performance, the company needs to make sure its customers don’t just walk away. Hence the steady decline in prices — though it’s also a natural result of streamlining models and improving their infrastructure.
GPT-3.5 Turbo also gets a new model release, 0125 (ie, today’s date) which includes “various improvements”, but apparently little that OpenAI thought worth mentioning. The last version was 0613 so it’s a bit surprising they don’t list more.
GPT-4 Turbo gets a new preview model for API usage, also 0125, which has an interesting fix:
This model completes tasks such as code generation more thoroughly than the previous preview model and is intended to reduce instances of “laziness” where the model does not complete a task.
Perhaps the model learned the wrong lessons from the programmers whose habits she swallowed and embodied! Maybe AI has already quietly stopped. Good for AI!
This is still in preview mode. But GPT-4 Turbo with vision (ie, GPT-4 V) will be in general availability “in the coming months.”
There are also a handful of new and improved text embedding models, which are more for the technical side. And the company also released a new version of its free moderation API — which identifies potentially harmful text. Look for version 007 if you are experimenting with using this to increase your moderation needs.
