This layer brings generation capabilities into your knowledge base. Whether you're building a chatbot, summarizer, or agent-driven workflow, Unbody's generative modules allow you to plug in large language models (LLMs) directly into your backend pipeline. These modules support use cases like text generation, structured JSON output, and (soon) image generation—enabling intelligent responses and content creation directly from your data.

Built-in Modules

Generative Text

Allows generation of freeform or conversational text from models based on prompt and context.

OpenAI

OpenAI language models available for general-purpose and instruction-following generation.

modules

Module Name	Module Description	Open Source Status	Cloud Status
gpt-4o-mini	Fast, optimized variant of GPT-4o for lightweight use cases	`Alpha`	`Available`
gpt-4o	Multi-modal, high-performance model for advanced generation	`Alpha`	`Available`
o1-mini	Smaller model for efficient, cost-effective tasks	`Alpha`	`Available`
o1	General-purpose instruction-following model	`Alpha`	`Available`
o3-mini	Compact model optimized for simple generation	`Alpha`	`Available`
gpt-4-turbo	High-efficiency version of GPT-4 with strong reasoning skills	`Alpha`	`Available`
gpt-4	Full-scale model offering advanced reasoning and language generation	`Alpha`	`Available`

Cohere

Support for full range of Cohere models.

modules

Module Name	Module Description	Open Source Status	Cloud Status
all available models	Placeholder for full Cohere model catalog	`Icebox`	`Available`

Mistral

Support for Mistral’s instruction-tuned models.

modules

Module Name	Module Description	Open Source Status	Cloud Status
all available models	Placeholder for full Mistral model set	`Icebox`	`Available`

Google AI

Future support for Gemini and other Google LLMs.