Requirements
System requirements and recommended LLMs for running Junet
Hardware
- CPU: 4 cores minimum
- RAM: 6 GB minimum
- Storage: 50 GB minimum
- Note: Storage requirements may increase depending on the number and size of embedded Agent data.
Operating System
- Operating System: Debian, Ubuntu, Red Hat Enterprise Linux
Ensure your system meets these minimum requirements for optimal performance. Higher specifications may be needed for production environments with multiple hundred users.
Recommended LLMs
We recommend the following Large Language Models for use with Junet. All models listed below have been tested and verified to work with Junet's requirements.
Required Features:
- Tooling/Function Calling support
- Vision capabilities
Models without these features will not work properly with Junet.
Self-Hosting LLMs:
If you plan to self-host any of the open-source models, please note:
- GPU Acceleration Required - Running LLMs locally requires a powerful GPU with sufficient VRAM
- Minimum GPU VRAM: 8-12 GB for smaller models (7-8B parameters)
- Recommended GPU VRAM: 16-24 GB for medium models (>24B parameters)
- High-end Datacenter GPU VRAM: 40+ GB for larger models (70B+ parameters)
Without adequate GPU hardware, model inference will be extremely slow or impossible. Commercial API-based models (ChatGPT, Claude) do not require local GPU resources.
Supported Models
| Name | Type | Status | Context Size | Params (B) |
|---|---|---|---|---|
You are free to experiment with other models not listed here. Just ensure they support both tooling/function calling and vision capabilities.
Testing Your Setup
After setting up your system, verify that:
- Your hardware meets the minimum specifications
- Your chosen LLM supports both tooling and vision
- All dependencies are properly installed
- The system has adequate storage for model weights
Start with a smaller model like Mistral-Small 3.2 for testing before deploying larger models in production.