Model Management Commands

VritraAI provides comprehensive AI model management capabilities. You can list, search, view, and switch between different AI models from various providers. The available models depend on your current API base (OpenRouter or Gemini).

Note: Available models change based on your current API base. Use api_base show to see which provider you're using, and api_base openrouter or api_base gemini to switch providers.

List All Models

model

List all available AI models organized by provider. Shows model IDs, display names, categories, descriptions, and model paths. The currently active model is marked with a checkmark.

Features:

  • Shows all models for current API base
  • Organized by provider
  • Displays model ID, display name, category, description, and path
  • Marks current active model
  • Shows usage examples and quick categories

Examples:

# List all available models
model

model list

Detailed listing of all available models with categories and descriptions. Provides comprehensive information about each model including provider, category, and full model path.

Features:

  • Detailed model information
  • Organized by provider
  • Shows model ID, display name, category, description, and path
  • Marks current active model

Examples:

# Detailed model listing
model list

View Current Model

model current

Show detailed information about the currently active model, including model ID, display name, provider, category, description, and internal path.

Features:

  • Shows current model details
  • Displays model ID, display name, provider, category, and path
  • Shows model description
  • Indicates if model is ready to use

Examples:

# Show current model
model current

Note: If no model is configured, the command will show a message and suggest using model list to see available models.

Switch Models

model set <id>

Switch to a specific AI model by its ID. The model selection is saved permanently and persists across sessions. When switching providers, the last used model for each provider is preserved.

Features:

  • Switch models by short ID (e.g., ds1, ll1, gf1)
  • Permanent configuration saving
  • Shows old and new model information
  • Provider-specific model preservation
  • Similar model suggestions if ID not found

Parameters:

  • id - Model ID (e.g., ds1, ll1, gf1, ms1)

Examples:

# Switch to DeepSeek Chat v3
model set ds1

# Switch to LLaMA 3.3 70B
model set ll1

# Switch to Gemini 2.0 Flash
model set gf1

# Switch to Mistral 7B
model set ms1

Tip: Use model list first to see all available model IDs for your current API base.

Search Models

model search <term>

Search for models by provider name, category, display name, or description. Useful for finding models that match specific criteria.

Features:

  • Search by provider (e.g., "deepseek", "google", "meta")
  • Search by category (e.g., "chat", "code", "large", "reasoning")
  • Search by keywords (e.g., "instruct", "turbo", "multimodal")
  • Shows matching models with full details
  • Marks current active model

Search Terms:

  • Provider names: deepseek, google, meta, mistral, openai, qwen
  • Categories: chat, code, large, medium, small, reasoning, multimodal, pro
  • Keywords: instruct, turbo, prover, flash, latest

Examples:

# Search for DeepSeek models
model search deepseek

# Search for coding models
model search code

# Search for large models
model search large

# Search for reasoning models
model search reasoning

# Search for multimodal models
model search multimodal

View Model Details

model view <id>

View detailed information about a specific model by its ID. Shows all model properties and indicates if it's currently active.

Features:

  • Detailed model information
  • Shows model ID, display name, provider, category, and path
  • Displays model description
  • Indicates if model is currently active
  • Provides command to switch to the model

Parameters:

  • id - Model ID to view

Examples:

# View DeepSeek Chat v3 details
model view ds1

# View LLaMA 3.3 70B details
model view ll1

# View Gemini 2.0 Flash details
model view gf1

Available Models

VritraAI supports models from multiple providers. The available models depend on your current API base:

Gemini Models (Google)

When using Gemini API base, the following models are available:

  • gf1 - Gemini 2.0 Flash (default) - Next-generation Gemini Flash model
  • gf2 - Gemini 2.5 Flash - Latest Gemini Flash with enhanced capabilities
  • gf3 - Gemini 2.5 Flash Lite - Lightweight and efficient Gemini Flash model
  • gf4 - Gemini Flash Latest - Latest Gemini Flash model with cutting-edge features
  • gp1 - Gemini 2.5 Pro - Premium Gemini model with advanced reasoning

OpenRouter Models

When using OpenRouter API base, the following models are available:

DeepSeek Models:

  • ds1 - DeepSeek Chat v3.0324 - Latest DeepSeek chat model with enhanced capabilities
  • ds2 - DeepSeek Chat - Standard DeepSeek conversational model
  • ds3 - DeepSeek Prover v2 - Mathematical reasoning and theorem proving

Meta LLaMA Models:

  • ll1 - LLaMA 3.3 70B Instruct - Latest LLaMA 3.3 with 70B parameters
  • ll2 - LLaMA 3.1 405B Instruct - Massive 405B parameter instruction-tuned model
  • ll3 - LLaMA 3.1 405B Base - 405B parameter base model for fine-tuning
  • ll4 - LLaMA 3.2 3B Instruct - Lightweight 3B parameter instruction model
  • ll5 - LLaMA 3 8B Instruct - Balanced 8B parameter instruction model
  • ll6 - LLaMA 3.1 8B Instruct - Updated 8B parameter instruction model

Mistral AI Models:

  • ms1 - Mistral 7B Instruct - Efficient 7B parameter instruction model
  • ms2 - Mistral Nemo - Advanced Mistral model with enhanced capabilities
  • ms3 - Pixtral 12B - Multimodal model with vision capabilities

Google Models:

  • gm1 - Gemma 3 27B IT - Google's 27B parameter instruction-tuned model
  • gm2 - Gemma 3 4B IT - Lightweight 4B parameter instruction model

OpenAI Models:

  • o1 - GPT-3.5 Turbo - Fast and efficient OpenAI model
  • o2 - GPT OSS 20B - Open source 20B parameter GPT model

Specialized Models:

  • qw1 - Qwen 3 Coder - Specialized coding and programming assistant
  • z1 - GLM 4.5 Air - Lightweight general language model

Model Categories

Models are organized into categories:

  • Chat: General conversational models
  • Code: Specialized coding and programming models
  • Reasoning: Models optimized for mathematical reasoning and problem-solving
  • Small: Lightweight models (3B-4B parameters)
  • Medium: Balanced models (7B-8B parameters)
  • Large: Powerful models (20B-70B parameters)
  • Ultra Large: Massive models (405B+ parameters)
  • Multimodal: Models with vision and text capabilities
  • Pro: Premium models with advanced capabilities

Best Practices

  • Choose by task: Use code models for programming, reasoning models for math, chat models for general conversation
  • Consider size: Smaller models are faster and cheaper, larger models are more capable
  • Check API base: Make sure you're using the correct API base for the models you want
  • Use search: Use model search to find models by category or provider
  • View details: Use model view to see full information before switching
  • Test models: Try different models to find the best one for your use case

💡 Pro Tip: Model selection is saved permanently. When you switch API bases, VritraAI remembers your last used model for each provider and automatically switches back when you return to that provider.