Like with any technology, there is constant development done to ensure improvement in the product/service. Similarly, GPT-4 is an improved version of GPT-3.5. There are no differences in the use case but, the responses generated by GPT-4 would be more accurate. Furthermore, the context in GPT-4 is 8,000 tokens, compared to the 4,000 tokens in GPT-3.5. This means that GPT-4 has more potential to read longer, detailed inputs and/or provide more detailed responses. However, for most requests, the 4,000 tokens are sufficient.
Copyright @2024 The University of Hong Kong. All Rights Reserved.