Article

Explore our curated collection of AI articles covering artificial intelligence, machine learning, AI tools and more

132 Article•Page 1 / 15

MedSAM3: Delving Deep into Specialized Areas Involving Medical Concepts

MedSAM3 delves into fine-grained domains involving medical concepts and supports Prompt Concept Segmentation (PCS) across multiple medical imaging modalities, including X-ray, MRI, ultrasound, CT, OCT, fundus, dermatoscopy, histopathology, and video.

Open source

SteadyDancer: Coherent and Smooth Human Image Animation with First Frame Preservation

SteadyDancer generates coordinated and coherent human image animations with first-frame preservation, ensuring consistent animation quality while maintaining robust first-frame retention and precise motion control.

Open source

Z-Image: Alibaba's Open-Source Image Generation Model

Z-Image is an open-source 6B-parameter image generation model from Alibaba, featuring Turbo (8 NFEs, sub-second latency), Base (for fine-tuning), and Edit variants for precise image editing with bi...

Open source

Dia2: An Open-Source Streaming Conversational TTS Model

Dia2 is an open-source streaming conversational TTS model that generates audio without requiring the full text upfront—audio generation can begin after just a few words are input. It enables real-time, natural conversation by conditioning the output on the incoming text.

Open source

FLUX.2 [dev]: Open-source text-to-image generation, editing, and composition model

FLUX.2 [dev], an open-source model for generating, editing, and composing images from text instructions, produces outputs suitable for personal, scientific, and commercial use.

Open source

HunyuanOCR: Tencent's Open-Source End-to-End OCR Expert VLM

HunyuanOCR, an open-source end-to-end OCR expert VLM from Tencent, adopts the native multimodal architecture of Hunyuan.

Open source

TiViBench: A Video Reasoning Benchmark for Video Generation Models

TiViBench: A Video Thinking Reasoning Benchmark for Video Generation Models, designed as a hierarchical benchmark to evaluate the reasoning capabilities of image-to-video (I2V) generation models.

Open source

MiMo-Embodied, Xiaomi's Open-Source Vision-Language Model

MiMo-Embodied, a powerful cross-embodied vision-language model that shows state-of-the-art performance in both autonomous driving and embodied AI tasks