What’s Updated

The available models have been updated to the following lineup:

• Qwen3-VL-235B-A22B-Instruct-FP8
• Qwen3.6-35B-A3B
• Gemma-4-31B-it
• DeepSeek-V4-Flash
• GPT-OSS-120b

Note: The previous models (gemma-3-27b-it and deepseek-r1:32b) have been replaced by newer versions for improved capability and user experience.

How to Select a Model in GenAI WebUI (Trial)

1. Open the GenAI service (https://genai.um.edu.mo/), and select GenAI WebUI (Trial) at the end.
2. On the top of the interface, open the model selector.
3. Choose the model that best fits your task.
4. Enter your prompt and submit.

Recommended Usage

• Qwen3-VL-235B-A22B-Instruct-FP8
  Best for multimodal tasks (e.g., understanding images, image‑based Q&A, vision + text workflows). Use this when you need the model to “see” and respond.

• Qwen3.6-35B-A3B 
  Best for more complex problem‑solving and multi‑step tasks, such as structured analysis, planning, troubleshooting, coding logic, and tasks that benefit from deeper reasoning.

• Gemma-4-31B-it
  Best for instruction-following daily tasks like rewriting, summarization, outlining, translation, formatting, and producing clean structured outputs quickly.

• DeepSeek-V4-Flash
  Best for fast responses and high‑throughput usage: quick Q&A, short drafting, iterative edits, and rapid back‑and‑forth refinement.

• GPT-OSS-120b
  Best for advanced reasoning and higher‑difficulty requests, including long/complex prompts, deeper analysis, and cases where quality matters more than speed.