Mu

GLM-OCR is a multimodal optical character recognition model and open source repository that provides accurate, efficient, and comprehensive document understanding by combining text and visual modalities into a unified encoder–decoder architecture derived from the GLM-V family. Built with a visual encoder pre-trained on large-scale image–text data and a lightweight cross-modal connector feeding into a GLM-0.5B language decoder, the model supports layout detection, parallel region recognition, and structured output for text, tables, formulas, and complicated real-world document formats. It introduces Multi-Token Prediction (MTP) loss and stable full-task reinforcement learning to improve training efficiency, recognition accuracy, and generalization, achieving state-of-the-art benchmarks on major document understanding tasks.

UNI-1 is a multimodal artificial intelligence model developed by Luma AI that unifies visual generation and reasoning capabilities within a single architecture, representing a step toward multimodal general intelligence. It was designed to overcome the limitations of traditional AI pipelines, where language models, image generators, and other systems operate independently without shared reasoning. UNI-1 integrates these capabilities so that language, visual understanding, and image generation work together inside one system, allowing the model to reason about scenes, interpret instructions, and generate visual outputs that follow logical and spatial constraints. At its core, UNI-1 is a decoder-only autoregressive transformer that processes text and images as a single interleaved sequence of tokens, enabling the model to treat language and visual information within the same computational framework rather than through separate encoders.