Information and links to come. Templates and guides are actively maintained and will be updated on an ongoing basis to reflect current standards, supported models, and version compatibility. Links will be provided, along with several “beginner” guides of how to log on, create your first model, upload to the knowledge base, etc.
The AI Lab LLM branch provides three chatbot interface instances designed to support research, education, and administrative use cases.
All interfaces require AUID authentication and have different availability scopes:
Note: LabBot and EduBot remain in beta. As such, they are subject to ongoing changes, including updates, potential model renaming, model substitutions, and feature adjustments. Certain functionalities - such as Channels and Notes - may still be unstable or behave inconsistently in practice.
Please note that EduBot is currently undergoing maintenance, which may cause document uploads and retrieval to be unstable. If your customized chatbot is experiencing difficulties retrieving a file, please try removing the file from the knowledge base and uploading it again.
Good at handling longer chats, large documents, and complicated instructions without losing track of context. Works well for research, coding (especially front-end), debugging, summarizing dense material, and tasks where the model needs to follow many constraints at once.
See https://artificialanalysis.ai/models/kimi-k2-6 for Intelligence and Performance Analysis.
Can spend additional time working through complex tasks before answering. Recommended for research, coding, analysis, and work that requires careful reasoning.
Prioritizes speed and responsiveness while maintaining Kimi's core strengths. Recommended for quick questions, summaries, drafting, and everyday use.
Technical specifications from https://artificialanalysis.ai/models/kimi-k2-6
| Reasoning | Yes |
| Input modality | Supports: text, image |
| Output modality | Supports: text |
| Context window | 256k (ca. 384 A4 pages of size 12 Arial font) |
| Total parameters | 1000B (1T) |
| Active parameters | 32B |
| Model weights | Hugging Face |
| Precision | NVFP4 quantized |
Good at breaking larger tasks into steps and working through them methodically. Useful for coding, editing documents, organizing information, spreadsheet-style work, and workflows where the chatbot model needs to plan before answering.
See https://artificialanalysis.ai/models/minimax-m3 for Intelligence and Performance Analysis.
Uses adaptive reasoning, allowing the model to spend extra time on complex tasks when beneficial while responding quickly to simpler requests. Recommended for most users and most tasks.
Prioritizes speed and responsiveness by minimizing additional reasoning. Recommended for quick questions, summaries, drafting, and everyday interactions.
Technical specifications from https://artificialanalysis.ai/models/minimax-m3
| Reasoning | Yes |
| Input modality | Supports: text, image |
| Output modality | Supports: text |
| Context window | 1M (ca. 1500 A4 pages of size 12 Arial font) |
| Total parameters | 428B |
| Active parameters | 23B |
| Model weights | Hugging Face |
| Precision | NVFP4 quantized |
Very strong for long coding tasks, agentic engineering, and workflows that need sustained reasoning over a large amount of context. Useful for projects where the chatbot needs to plan, test, revise, and continue making progress over many steps.
See https://artificialanalysis.ai/models/glm-5-2 for Intelligence and Performance Analysis.
Uses the highest level of reasoning available and may spend additional time exploring alternatives and verifying its approach. Best suited for difficult coding, deep research, advanced analysis, and complex multi-step tasks.
Uses enhanced reasoning to work through more complex problems before answering. Recommended for analysis, coding, research, and tasks requiring careful evaluation.
Optimized for fast responses with no additional reasoning. Best for quick questions, summaries, drafting, and general chat.
Technical specifications from https://huggingface.co/zai-org/GLM-5.2
| Reasoning | Yes |
| Input modality | Supports: text |
| Output modality | Supports: text |
| Context window | 1M (ca. 1500 A4 pages of size 12 Arial font) |
| Total parameters | 753B |
| Active parameters | 40B |
| Model weights | Hugging Face |
| Precision | NVFP4 quantized |
