IBM Granite

AI model · Multimodal model · Text → Text by IBM

Family of open, trusted AI models for business

IBM Granite is a family of open, trusted AI models designed for business use, offering efficient language, vision, speech, and guardrail models that organizations can customize and deploy. The models serve enterprises seeking to build AI applications with transparent, governable foundations. IBM provides these models through its watsonx platform and other channels, allowing businesses to adapt them for specific domains while maintaining control over data and deployment. The family emphasizes openness and trust, with models that can be fine-tuned for enterprise requirements. Granite models are

Languages
English, Spanish
Modalities
Text, image, speech (varies by specific model)
License
Open
Access
Via watsonx platform, API, and download for customization
Languages
Multiple including English, Spanish

Overview

IBM Granite is a family of open, trusted AI models designed for business use, offering efficient language, vision, speech, and guardrail models that organizations can customize and deploy. The models serve enterprises seeking to build AI applications with transparent, governable foundations. IBM provides these models through its watsonx platform and other channels, allowing businesses to adapt them for specific domains while maintaining control over data and deployment. The family emphasizes openness and trust, with models that can be fine-tuned for enterprise requirements. Granite models are used across IBM's own products and by clients for tasks ranging from code generation to content analysis.

Language models

  • Text generation: Efficient language models for generating and understanding text across business domains.
  • Code generation: Models optimized for software development tasks including code completion and modernization.
  • Multilingual support: Supports multiple languages including Spanish for global enterprise applications.

Vision models

  • Image understanding: Models for visual analysis and computer vision tasks in enterprise workflows.

Speech models

  • Speech recognition: Models for converting spoken language to text in business applications.
  • Speech synthesis: Models for generating natural-sounding speech from text inputs.

Guardrail models

  • Safety and moderation: Models designed to enforce safety policies and moderate content in AI applications.
  • Compliance support: Guardrails that help ensure AI outputs meet enterprise governance requirements.

Specifications

Modalities
Text, image, speech (varies by specific model)
License
Open
Access
Via watsonx platform, API, and download for customization
Customization
Models can be fine-tuned and deployed per enterprise requirements
Languages
Multiple including English, Spanish

Added by AntSpace from public sources (ibm.com).

More AI models by IBM