Skip to main content

Module models

Module models 

Source
Expand description

Pre-configured model definitions

This module contains model structs for various LLM providers. Each model implements the Model trait and provider-specific traits like BedrockModel or AnthropicModel.

Models are organized by vendor:

  • claude - Anthropic Claude models
  • llama - Meta Llama models
  • nova - Amazon Nova models
  • mistral - Mistral AI models
  • cohere - Cohere models
  • qwen - Alibaba Qwen models
  • google - Google models
  • deepseek - DeepSeek models
  • kimi - Moonshot Kimi models

Structsยง

Claude3_7Sonnet
Claude 3.7 Sonnet - Latest Claude 3.x with improved reasoning
ClaudeHaiku4_5
Claude Haiku 4.5 - Fast, efficient model for high-throughput tasks
ClaudeOpus4
Claude Opus 4 - High capability reasoning model
ClaudeOpus4_1
Claude Opus 4.1 - Advanced reasoning model
ClaudeOpus4_5
Claude Opus 4.5 - High-capability reasoning and creative writing model
ClaudeOpus4_6
Claude Opus 4.6 - Flagship Claude model with 128K output
ClaudeSonnet4
Claude Sonnet 4 - Balanced performance and cost
ClaudeSonnet4_5
Claude Sonnet 4.5 - Latest Sonnet with improved capabilities
CohereCommandRPlus
Command R+ - Enterprise RAG and multi-step tool use model
DeepSeekR1
DeepSeek R1 - Reasoning-focused model
DeepSeekV3_1
DeepSeek V3.1 - General purpose model
DeepSeekV3_2
DeepSeek V3.2 - Updated general purpose model
Gemma3_4B
Gemma 3 4B - Compact open model from Google
Gemma3_12B
Gemma 3 12B - Mid-size open model from Google
Gemma3_27B
Gemma 3 27B - Open multimodal model from Google
KimiK2Thinking
Kimi K2 Thinking - Reasoning-enhanced model from Moonshot AI
KimiK2_5
Kimi K2.5 - Next-gen model from Moonshot AI
Llama3_1_8B
Llama 3.1 8B Instruct - Efficient general purpose model
Llama3_1_70B
Llama 3.1 70B Instruct - High capability model
Llama3_1_405B
Llama 3.1 405B Instruct - Largest open-weights model
Llama3_2_1B
Llama 3.2 1B Instruct - Lightweight model for edge deployment
Llama3_2_3B
Llama 3.2 3B Instruct - Efficient small model
Llama3_2_11B
Llama 3.2 11B Instruct - Medium multimodal model
Llama3_2_90B
Llama 3.2 90B Instruct - Large multimodal model
Llama3_3_70B
Llama 3.3 70B Instruct - Latest Llama 3.x flagship
Llama4Maverick17B
Llama 4 Maverick 17B - Larger MoE model with 1M context
Llama4Scout17B
Llama 4 Scout 17B - Efficient MoE model with 10M context
MagistralSmall
Magistral Small - Efficient 24B reasoning model with vision
Ministral3B
Ministral 3B - Compact 3B instruction model
Ministral8B
Ministral 8B - Efficient 8B instruction model
Ministral14B
Ministral 14B - Mid-size 14B instruction model
MistralLarge3
Mistral Large 3 - Flagship 675B MoE model with 41B active parameters
Nova2Lite
Nova 2 Lite - Fast reasoning model with extended thinking support
Nova2Sonic
Nova 2 Sonic - Next-gen Nova model with 1M context
NovaLite
Nova Lite - Multimodal model for image, video, and text
NovaMicro
Nova Micro - Lightweight, text-only model for simple tasks
NovaPremier
Nova Premier - Highest capability Nova model with 1M context
NovaPro
Nova Pro - Balanced multimodal model
PixtralLarge
Pixtral Large - Vision-capable large model
Qwen3Coder30B
Qwen3 Coder 30B - Compact coding-focused MoE model
Qwen3Coder480B
Qwen3 Coder 480B - Large coding-focused MoE model
Qwen3Next80B
Qwen3 Next 80B - Next-gen MoE model
Qwen3VL235B
Qwen3 VL 235B - Vision-language MoE model
Qwen3_32B
Qwen3 32B - Dense 32B model
Qwen3_235B
Qwen3 235B - Large MoE model with 22B active parameters
VoxtralMini3B
Voxtral Mini 3B - Speech and text input model
VoxtralSmall24B
Voxtral Small 24B - Speech and text input model