Model Configuration

Configure orchestration and domain models

Logged in as: {{ user.name }}
⚠️ Warning: Changes to model configuration require restarting the application to take effect. Make sure to save your configuration before restarting.

🎯 Orchestration Configuration

The orchestrator uses TinyBERT for intent classification and Phi-2 for complex reasoning.
Small model for fast intent classification
Model for complex query analysis
Threshold for triggering reasoning (0.0-1.0)
Use Phi-2 for complex queries

🤖 Domain Model Configuration

Each domain can use a different specialized model. Changes are saved to the configuration file.

📦 Available Models

  • deepseek-v3.1:671b-cloud Large
  • kimi-k2-thinking:cloud Reasoning
  • qwen3-coder:480b-cloud Coding
  • gpt-oss:120b-cloud General
  • qwen3-vl:235b-instruct-cloud Vision
E.g., llama3:8b, mistral:7b
💡 Configuration Tips:
Save Configuration: Saves your changes to mdsa_config.json
Reset to Defaults: Restores factory default settings
Reload from File: Reloads domains from mdsa_config.json without restarting the dashboard (use this after manually editing the config file)