GPT-5.4-mini (2026-03-17)
OpenAI released GPT-5.4 mini on 17 March 2026. GPT‑5.4 mini significantly improves upon GPT‑5 mini in coding, reasoning, multimodal understanding, and tool use, while running more than 2x faster. It also approaches the performance of the larger GPT‑5.4 model on SWE-Bench Pro and OS World-Verified. The GPT-5.4-mini model’s training data covers information up to August 2025.
The PolyU Gen AI app released GPT-5.4 mini (2026-03-17) on 1 May 2026.
The GPT-5.4 mini model, hosted on Microsoft Azure Cloud and accessed via the PolyU GenAI app, consumes 0.75 credits per 1K tokens for text or image input and 4.5 credits per 1K output tokens, including chain-of-thought reasoning.
Staff and students can leverage the “reasoning effort” parameter to control the extent of reasoning (inference looping) performed by the GPT-5.4 mini model. Additionally, the “Reasoning summary” feature is enabled for all users to display summaries of the model's chain-of -thought reasoning.
For more details on GPT-5.4 mini, please refer to Introducing GPT‑5.4 mini and nano, Introducing OpenAI’s GPT-5.4 mini and GPT-5.4 nano for low-latency AI and GPT-5.4-mini on PolyU GenAI.
Kimi-K2.5
Moonshot AI released Kimi K2.5 on 27 January 2026. Kimi K2.5, Moonshot AI’s most powerful open-source model to date, was built on Kimi K2 with continued pretraining on approximately 15T mixed visual and text tokens. It seamlessly integrates vision and language understanding with advanced agentic capabilities, offering instant and thinking modes, as well as both conversational and agentic paradigms.
The PolyU GenAI app introduced Kimi K2.5 on 1 May 2026.
The Kimi K2.5 model, hosted on Alibaba Cloud Bailian and accessible via the PolyU GenAI app, consumes 0.56 credits per 1K tokens for text or image input and 2.94 credits per 1K tokens for text, including chain-of-thought reasoning.
You may choose to enable or disable the thinking mode of Kimi K2.5.
- Thinking Mode: In this mode, the model takes time to reason step by step before delivering the final answer. This is ideal for complex problems that require deeper thought.
- Non-Thinking Mode: Here, the model provides quick, near-instant responses, making it suitable for simpler questions where speed is more important than depth.
For more details on Kimi K2.5, please refer to the Kimi K2.5 Technical Blog, Technical Report of Kimi K2.5 and Kimi-K2.5 on PolyU GenAI.
Qwen3.6-Plus (2026-04-02)
Alibaba Cloud released Qwen3.6-Plus in April 2026. Following the release of the Qwen3.5 series in February, this version represents a major capability upgrade over its predecessor. Most notably, Qwen3.6-Plus has drastically enhanced agentic coding capabilities.
The Qwen3.6-Plus model introduces a hybrid approach to problem-solving, supporting two modes:
- Thinking Mode: The model takes time to reason step by step before delivering the final answer. This is ideal for complex problems requiring deeper thought.
- Non-Thinking Mode: The model provides quick, near-instant responses, making it suitable for simpler questions where speed is prioritized over depth.
The PolyU GenAI app released Qwen3.6-Plus (2026-04-02) on 1 May 2026, replacing Qwen3-Max (2026-01-23).
Qwen3.6-Plus (2026-04-02), hosted on Alibaba Cloud Bailian and accessed via the PolyU GenAI app, consumes credits as follows:
- For input tokens up to 256K: 0.28 credits per 1K tokens for text or image input, and 1.68 credits per 1K output tokens for text, including chain-of-thought tokens.
- For input tokens exceeding 256K: 1.12 credits per 1K tokens for text or image input, and 6.72 credits per 1K output tokens for text, including chain-of-thought tokens.
For details on Qwen3.6-Plus, please refer to the Qwen3.6-Plus Official Blog and Qwen3.6-Plus on PolyU GenAI.
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.
