Latest News

Note: The usage guide for the Virtual Try-On Application is now available. This project implements a virtual try‑on application using Flask, Twilio, and Gradio. Users can upload images to test different clothing combinations. The project is open‑source, making it easy for developers to implement personalized try‑on features in their own systems.

Mochi: Commercially Available! The Largest Open-Source Video Generation Model to Date Arrives!

Recently, Genmo AI released its latest video generation model, the Mochi 1 preview version, as open source. Mochi is an advanced open video generation model that delivers high-fidelity motion and strong prompt adherence. Mochi 1 markedly narrows the gap between open video generation models and proprietary alternatives. It is released under the Apache 2.0 license, permitting free commercial use for both individuals and enterprises. A 480p base model is already available on HuggingFace, and the Mochi 1 HD version is slated for release by the end of the year. Additionally, Genmo AI announced the completion of a $28.4 million Series A financing round led by NEA.

Super Popular! MimicTalk – Train Your Digital Human in 15 Minutes

Train a high‑quality, personalized digital human in just 15 minutes! MimicTalk is a 3D digital‑human generation project jointly developed by Zhejiang University and ByteDance, leveraging Neural Radiance Fields (NeRF) technology to create personalized, lifelike 3D speaking faces within 15 minutes. Compared with traditional methods, MimicTalk significantly improves generation efficiency and expressiveness, producing videos that are more realistic and vivid.

Hunyuan3D-1.0 – Tencent's 3D Generation Model Supporting Text-to-3D and Image-to-3D

Hunyuan3D-1.0 is a powerful 3D generation model released by Tencent that supports both text and image inputs, enabling rapid creation of high‑quality 3D assets. It employs a two‑stage generation approach: first, a multi‑view diffusion model produces multi‑view RGB images; then, a transformer‑based sparse‑view large‑scale reconstruction model converts these images into a 3D model. The model is available in a lightweight version for quick modeling and a standard version that delivers higher‑quality 3D results.

Digital Human

Digital Marketing

Ichigo – Open‑Source Multimodal AI Voice Assistant that Processes Interleaved Speech and Text Sequences in Real Time

Ichigo is an open‑source multimodal AI voice assistant that leverages a hybrid modality model to handle interleaved speech and text streams instantly. By directly quantizing speech into discrete tokens and employing a unified transformer architecture that simultaneously processes audio and text, Ichigo achieves cross‑modal joint inference and generation. This design boosts processing speed and efficiency, delivering a latency of just 111 ms—substantially faster than existing solutions—and providing a near‑real‑time voice interaction experience.

Large Model

A Professional Guide to Improving the Accuracy of GPT-Generated JSON Data: How to Make AI Produce 100% Perfect JSON

This article introduces how to improve the accuracy of GPT-generated JSON format data, ensuring AI output fully meets project requirements. The content includes three major steps: precise prompt design, dynamic constraint decoding control, and post-processing correction, progressively optimizing the generation process and significantly enhancing the structural accuracy of JSON data. It is suitable for users who need to handle complex data streams and large-scale datasets; these methods help developers achieve efficient and precise data output in AI projects, easily tackling data processing challenges.

Microsoft OmniParser Open-Source UI Parser: An Automation Powerhouse That Outperforms GPT-4V!

Microsoft has officially released the OmniParser open-source UI parser, demonstrating outstanding performance in screen parsing and comprehension, even surpassing GPT-4V in benchmark tests. This tool can convert UI screenshots into structured formats, significantly enhancing the screen understanding capabilities of automation tools and AI assistants.

Recommended

Product Transformation: Founder Builds Demo in 48 Hours, Company Valuation Soars to $650 Million in Two Months

Casetext's successful AI transformation showcases the huge potential of AI products in vertical markets. Founder Jake Heller, after experiencing GPT-4, built a demo of the legal AI assistant CoCounsel in just 48 hours, and within two months raised the company's valuation to $650 million, eventually being acquired by Thomson Reuters. Heller detailed how the team leveraged test‑driven development and prompt engineering to fine‑tune AI output accuracy, ensuring the product is suitable for critical legal tasks, and noted that the success of vertical AI products depends on unique data, business logic, and engineering design. This case not only validates the massive business opportunity for AI in the legal sector, but also demonstrates that AI transformation can achieve product‑market fit and rapid growth by quickly responding to market changes.

Recommended

Musk: Brain-Computer Interface Will Transform Treatment of Brain Disorders, Target Cost $5,000

At the 2024 Neurosurgery Physicians Conference, Elon Musk announced that Neuralink's brain-computer interface technology is expected to help address most brain disorders, with a future goal of reducing the device cost to $5,000. By capturing neural signals, this technology aims to treat conditions such as depression and Parkinson's disease, making brain disorder treatment more accessible and ushering in a new era of efficient and affordable healthcare.

A 17-Year-Old High School Student's Million-Dollar AI App: Is This the Dawn of a New Era for Independent Developers?

Seventeen‑year‑old high school student Zach generated a million dollars in revenue within four months by developing the weight‑management app Cal AI. Cal AI leverages image‑recognition technology to analyze food calories, enabling users to manage their weight scientifically. The app’s success stems from addressing a genuine need and employing an innovative social‑media distribution strategy. One of the team members, Brake, taught himself AI programming and distilled a growth formula based on uncovering demand, low‑cost promotion, and rapid validation. Cal AI’s triumph signals the rise of the “quick‑app” wave, where independent developers validate market demand and monetize through single‑function applications. This case showcases market opportunities for AI indie developers while highlighting the sharp market insight and effective promotion tactics required for success.

Large Models

Zhipu AI Launches Globally Leading Agent AutoGLM: Complete Phone Operations with a Single Sentence, Fully Liberating Hands

Zhipu AI recently unveiled its newest agent, AutoGLM, delivering the convenience of "one sentence to handle phone operations." Users simply voice their request, and AutoGLM automatically performs a variety of complex tasks on a smartphone or web interface—ordering food delivery, booking hotels, shopping, and more. The core technologies behind AutoGLM include a decoupled design for task planning and action execution, as well as a self‑learning framework, which make its operations more precise and flexible while gradually improving task completion rates. In addition, Zhipu AI released the emotional speech model GLM‑4‑Voice, which supports multiple emotional expressions, flexible output, and multilingual capabilities, providing a natural and fluent interactive experience. These two innovations offer users a brand‑new intelligent lifestyle.

Large Model