From foundation language models to unified multimodal perception and generation — open-source and built for everyone.
A Mixture-of-Experts large language model fully open-sourced by InclusionAI.
A reasoning-focused MoE language model open-sourced by InclusionAI.
An updated lightweight reasoning model with improved performance.
Novel chain-of-thought prompting methods for enhanced reasoning.
A unified reasoning framework covering general and spatial understanding.
Unified multimodal model processing images, text, audio, and video.
Updated lightweight omni model with improved multimodal capabilities.
Preview release of Ming-Lite-Omni with broad modality support.
Flash variant with sparse architecture for efficient multimodal processing.
Extends Ming-Lite-Omni with segmentation-as-editing capabilities.
Unified architecture enabling natural interaction across modalities.
Text-to-speech and audio generation with fine-grained control.
Speech language model unifying audio understanding and generation.
Unified vision model for joint image understanding and generation.