Article

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

17 Article•Page 1 / 2

xAI launches Grok 4 Fast, a multimodal reasoning model with a 2M context window

xAI launches Grok 4 Fast, a multimodal reasoning model with a 2M context window, available for free on Grok.com, iOS and Android apps, and OpenRouter.

Open source

Qwen-Image-Pruning, Based on Open-Source Qwen-Image

Qwen-Image-Pruning, based on the open-source Qwen-Image model, is a 13.3B-parameter model launched by OPPO.

Open source

Wan2.2-Animate: An Open-Source Unified Character Animation Model

Wan2.2-Animate is Wan's open-source unified model for high-fidelity character animation and replacement, featuring full-body motion and expression replication.

Open source

MiMo-Audio: Xiaomi's Open-Source Audio Language Model

MiMo-Audio, a speech-language model open-sourced by Xiaomi, achieves state-of-the-art performance on speech intelligence and audio understanding benchmark tests among open-source models, with pre-training data expanded to over 100 million hours.

Open source

Granite Docling: IBM's Open-Source Multimodal Document Processing Model

Granite Docling, an open-source multimodal document processing model from IBM, is designed for efficient document transformation and supports a variety of functionalities.

Open source

SAIL-VL2: ByteDance's Open-Source Vision-Language Model

SAIL-VL2, a vision-language model open-sourced by ByteDance, achieves comprehensive multimodal understanding and reasoning, excelling at 2B and 8B parameter scales.

Open source

Lucy Edit Dev: An Open-Source Video Editing Model

Lucy Edit Dev is an open-source video editing model that performs instruction-guided edits on videos using free-text prompts. For example, it can change clothing and accessories, modify characters, insert objects, or replace scenes while perfectly preserving motion and composition.

Open source

Qwen3-ASR-Toolkit:This ensures that words and sentences are not awkwardly cut off

Qwen3-ASR-Toolkit is an advanced, high-performance Python command-line toolkit for using the Qwen-ASR API (formerly known as Qwen3-ASR-Flash).

Open source

TimesFM: A Pretrained Time Series Foundation Model for Time Series Forecasting from Google Research

TimesFM is an open-source pre-trained foundation model for time series forecasting from Google Research. With 200M parameters—under 500M—it supports context lengths up to 16k.

Open source

Stable Part Diffusion 4D: Multi-View RGB and Kinematic Part Video Generation

Stable Part Diffusion 4D: A framework for multi-view RGB and kinematic part video generation, enabling paired RGB and motion-aware part videos from monocular input.

Information

GPT-5-Codex: OpenAI's Enhanced GPT-5 Version Optimized for Agent Coding in Codex

GPT-5-Codex is a version of GPT-5 developed by OpenAI, specifically optimized for agent-based coding in Codex, and has been trained on complex, real-world engineering tasks.