Search papers, labs, and topics across Lattice.
Dynin-Omni is introduced as the first masked-diffusion-based omnimodal foundation model, unifying text, image, speech, and video within a single architecture. This model formulates omnimodal modeling as masked diffusion over a shared discrete token space, enabling iterative refinement under bidirectional context, unlike autoregressive or compositional approaches. Evaluated across 19 multimodal benchmarks, Dynin-Omni achieves strong performance, outperforming existing open-source unified models and remaining competitive with modality-specific expert systems.
Masked diffusion unlocks a unified paradigm for any-to-any multimodal modeling, rivaling specialized models without modality-specific decoders.
We present Dynin-Omni, the first masked-diffusion-based omnimodal foundation model that unifies text, image, and speech understanding and generation, together with video understanding, within a single architecture. Unlike autoregressive unified models that serialize heterogeneous modalities, or compositional unified models that require orchestration with external modality-specific decoders, Dynin-Omni natively formulates omnimodal modeling as masked diffusion over a shared discrete token space, enabling iterative refinement under bidirectional context. Dynin-Omni adopts a multi-stage training strategy with model-merging-based modality expansion and omnimodal alignment. We evaluate Dynin-Omni across 19 multimodal benchmarks spanning language reasoning, image generation and editing, video understanding, and speech recognition and synthesis. Dynin-Omni achieves 87.6 on GSM8K, 1733.6 on MME-P, 61.4 on VideoMME, 0.87 on GenEval, and 2.1 WER on LibriSpeech test-clean, consistently outperforming existing open-source unified models while remaining competitive with strong modality-specific expert systems. These results demonstrate the potential of masked diffusion as a unified paradigm for any-to-any modeling, providing a flexible foundation for real-time omnimodal systems, unified cross-modal retrieval and generation, and embodied multimodal agents.
