Gemini 3 Pro (Preview)
Google Cloud released the Gemini 3 Pro Preview model in November 2025. Compared to Gemini 2.5 Pro, the Gemini 3 Pro Preview model demonstrates significant improvements in high-level reasoning, complex instruction following, agentic and tool use cases, as well as long context capabilities.
The PolyU Gen AI app released the Gemini 3 Pro Preview model on 1 February 2026.
The Gemini 3 Pro Preview, hosted on Google Cloud and accessed via the PolyU Gen AI app, will consume credits as follows:
- < = 200K input tokens: 2 credits per 1K tokens for text or image input, and 12 credits per 1K tokens for output text, including chain-of-thought tokens.
- > 200K input tokens: 4 credits per 1K tokens for text or image input, and 18 credits per 1K tokens for output text, including chain-of-thought tokens.
For details of Gemini 3 Pro Preview, please refer to Introduction to Gemini 3 Pro and the Gemini 3 Pro Model Card.
Gemini 3 Flash (Preview)
Google Cloud released the Gemini 3 Flash Preview on 17 December 2025. The Gemini 3 Flash Preview significantly outperforms Gemini 2.5 Pro across a range of benchmarks requiring enhanced reasoning and multimodal capabilities. The Gemini 3 Flash Preview model is designed to tackle agentic workflows with efficiency, low latency, and cost-effectiveness.
The PolyU Gen AI app released the Gemini 3 Flash Preview model on 1 February 2026.
Gemini 3 Flash, hosted on Google Cloud and accessed via the PolyU Gen AI app, will consume 0.5 credits per 1K tokens for text or image input and 3 credits per 1K tokens for output text, including chain-of-thought tokens.
For details of Gemini 3 Flash, please refer to Introduction to Gemini 3 Flash and the Gemini 3 Flash Model Card.
Deep Research Tool Updates
We are excited to announce updates to the Deep Research Tool on the PolyU Gen AI App, effective 1 February 2026:
- Upgrade of the Preview release of the Deep Research Tool from using the Gemini 2.5 Pro model → the Gemini 3 Pro (Preview) model as a judge, available to Academic, Teaching, & Clinical colleagues.
- Preview release of the Deep Research Tool using the Tencent HY 2.0 Think model as a judge, available to all staffs and students.
The in-house developed Deep Research Tool is designed to deliver a cost-effective research assistant that is secure, flexible and tailored to the unique needs of the academic community.
Unlike conventional Internet search tools that simply return a list of web links, the Deep Research Tool leverages advanced AI models to act as an intelligent research agent. The system can break down complex research queries into manageable subtasks, independently gather and analyse information from both the internet and academic databases, and synthesize these findings into a comprehensive, well-structured report complete with citations. This process ensures that users receive thorough and academically rigorous response to their research questions.
For further details on how to make the most of the Deep Research Tool, including tips on formulating effective research topics and understanding credit usage, please refer to PolyU_GenAI_Deep_Research_Introduction. You may also find a sample report in PolyU_GenAI_Deep_Research_GPT5.1_Example. We look forward to your feedback as we continue to refine and improve on Deep Research tool for the PolyU community.
If you need further information or assistance, please contact the IT HelpCentre (Tel: 2766 5900, WhatsApp/ WeChat: 6577 9669) or reach out to us via the IT Online ServiceDesk.
