We are thrilled to introduce our new service, GPT-4o, which will be accessible to HKU students and staff in the upcoming semester at HKU ChatGPT. Starting from 31 August 2024, HKU users could make use of this service and GPT-4o supports image analysis.
GPT-4o (“o” for “omni”) represents a significant advance in natural human-computer interaction. It allows users to input not only text but also images. This enhancement includes recognizing objects and understanding context in images. As a result, users can seamlessly embed images into their queries to allow AI to interpret and analyze visual data and provide more relevant results. Compared with GPT-4, GPT-4o uses fewer tokens for both prompts and responses.
For Example:
| Model | Tokens for Prompt | Tokens for Response |
| GPT-4 | 6 | 4 |
| GPT-4o | 1 | 1 |
Students and Staff could continue to access GPT-3.5 and GPT-4. Due to the upgrade of GPT-3.5, the new GPT-3.5 will provide faster response times, enhanced text generation quality, and lower costs on token usage.
For Example:
| Model | Tokens for Prompt | Tokens for Response |
| GPT-3.5 (Old) | 4 | 4 |
| GPT-3.5 (New) | 1 | 3 |