PolyU GenAI App models refresh on Sept 1st.

OpenAI released the GPT-5 and GPT-5 mini models in August 2025. The GPT-5 series of models outperform across the board, with major gains in coding, instruction following and long context understanding. This series of GPT-5 models also have new features of fine tuning in reasoning effort and verbosity to adjust the speed, quality and length of responses. The GPT-5 series models’ knowledge training covers data up to October 2024.

PolyU GenAI app released GPT-5 (2025-08-07) on 1 September 2025, replacing o3 (2025-04-16).
PolyU GenAI app released GPT-5-mini (2025-08-07) on 1 September 2025, replacing o4-mini (2025-04-16).

Staff and students can leverage the “reasoning effort” parameter to control the extent of reasoning (inference looping) required to be carried out by the GPT-5 model. Also, the “Reasoning summary” feature is enabled for all users to display summaries of the model's chain of thought reasoning.

The GPT-5 model hosted on MS Azure Cloud and accessed via the PolyU GenAI app consumes 1.25 credits per 1K tokens on text or image input and 10 credits per 1K tokens on output text, including chain-of-thought. The GPT-5-mini model hosted on MS Azure Cloud and accessed via the PolyU GenAI app consumes 0.25 credits per 1K tokens on text or image input and 2 credits per 1K tokens on output text, including chain-of-thought.

Google Cloud released the Gemini 2.5 Pro and Gemini 2.5 Flash models in June 2025. Gemini 2.5 Pro and Gemini 2.5 Flash are multimodal reasoning models that can comprehend vast datasets and information sources, including text, audio, images, video, and even entire code repositories. The Gemini 2.5 Pro and Gemini 2.5 Flash models’ knowledge covers training data up to January 2025.

Gemini 2.5 Pro and Gemini 2.5 Flash support grounding with the Google Search tool, where Gemini 2.5 Pro and Gemini 2.5 Flash include information from web searches to answer users’ questions.

PolyU GenAI app released the Gemini 2.5 Pro and Gemini 2.5 Flash models on 1 September, 2025.

Staffs and students can leverage the “reasoning effort” parameter to control the extent of reasoning (inference looping) required to be carried out by the Gemini 2.5 Pro and Gemini 2.5 Flash model. Also, the “Reasoning summary” feature is enabled for all users to display summaries of the model's chain-of-thought reasoning.

The Gemini 2.5 Pro model, hosted on Google Cloud and accessed via the PolyU GenAI app, consumes 2.5 credits per 1K tokens on text or image input and 15 credits per 1K tokens on output text, including chain-of-thought. The Gemini 2.5 Flash model, hosted on Google Cloud and accessed via the PolyU GenAI app, consumes 0.3 credits per 1K tokens on text or image input and 2.5 credits per 1K tokens on output text, including chain-of-thought.

Alibaba released the Qwen3-235B-A22B-Thinking-2507 model in July 2025. This model has improved thinking capability in both the quality and depth of reasoning, including logical reasoning, mathematics, science, coding, and academic benchmarks that typically require human expertise, with enhanced 256K long-context understanding capabilities. Unlike the hybrid-thinking mode supported in the previous Qwen3-235B-A22B model, this model only supports thinking mode.

The Qwen3-235B-A22B-Thinking-2507 reasoning model was released on PolyU Gen AI on 1 September 2025, for use by PolyU staff and students.

The use of the Qwen3-235B-A22B-Thinking-2507 open-source model will not consume the monthly usage credit entitlement.

FLUX.1 Kontext [dev] is a 12 billion parameter rectified flow transformer capable of editing images based on text instructions. This model was released by The Black Forest Team on 26 June 2025.

FLUX.1 Kontext [dev] focuses exclusively on editing tasks. The model enables iterative editing, excels at character preservation across a diverse set of scenes and environments, and allows both precise local and global edits. The model’s abilities include in-context image generation, localized image editing, image merging, and preserving prompt fidelity.

Staff and students can fine-tune prompt, guidance, seed and steps parameters under settings to control the effects of the output image.

The FLUX.1 Kontext [dev] model was released on PolyU GenAI on 1 September 2025, for use by PolyU staff and students.

PolyU GenAI app provides image(s)-to-image service for Flux.1 Kontext [dev] only. In other words, at least one image should be uploaded when using this model for image editing, and at most two images can be uploaded for image combining.

The use of the FLUX.1 Kontext [dev] open-source model will not consume the monthly usage credit entitlement.