Search “Video”

MiniMax responds to H3 integration with TPU: looking forward to open-source work on efficient video inference

MiniMax officially responded to SGLang adding MiniMax H3 to the TPU support list of SGLang-TorchTPU, and said it looks forward to more open-source work on efficient video inference. SGLang said that SGLang-TorchTPU was built jointly by SGLang, the Google TPU team, and the Meta PyTorch team, has been used for production inference at Meta, and supports models including Muse Spark Open, GLM-5.3 Flash, Kimi K3, Ling 3.1 Flash, and MiniMax H3.

English translation is being prepared.

Nikon Microphotography Contest Winner Stripped of Award Over AI-Generated Video

Nikon's 16th Small World in Motion microphotography competition has revoked the championship title of Dr. Ning Xu of Tsinghua University, whose work was deemed to violate the competition rules. Researchers pointed out that the cell structures in the video were abnormally sized and cilia appeared out of nowhere; Ian Donovan, a doctoral student at UT Southwestern, also found that the source image carried an AI-generation watermark. In a deleted LinkedIn post, Xu acknowledged using AI for post-processing but said AI was not used to generate the video itself; Nikon stated it will re-examine the rules and judging process for future competitions, and the championship was awarded instead to Nguyen Nam Nhat of Vietnam.

English translation is being prepared.

English translation is being prepared.

English translation is being prepared.

English translation is being prepared.

English translation is being prepared.

English translation is being prepared.

Wan3.0 ranks first in the new Artificial Analysis video generation benchmark

Wan3.0 ranks first overall in the newly launched Artificial Analysis AA-Video-T2V v2.0 text-to-video benchmark, and leads in 10 of 20 category leaderboards, including cartoon and anime styles and animation and gaming use cases, with a video cost of 12 USD per minute. The benchmark is based on more than 1000 prompts and over 68000 human preference votes from private evaluators in the US and UK, and uniformly judges models at 1080p high bitrate. Wan officially thanks the team, the evaluators, and the creators who use Wan.

Iluvatar CoreX Tianyao 300 Benchmarked Against H200: Training Throughput Reaches 1.33x That of H100

Iluvatar CoreX held its first live demonstration of the Tianyao 300 at a media open day on October 28, claiming that in comparative tests on large language models such as GLM-5.3 as well as text-to-image and text-to-video scenarios, it outperformed the H200 overall. On the training side, in an 8-GPU full-system FP8-precision demonstration based on Qwen3-32B, the Tianyao 300 achieved per-GPU effective throughput 1.33 times that of the H100. The Tianyao 300 was unveiled at this year's World Artificial Intelligence Conference in July, features 144GB of memory, and its IXL144 supernode integrates 144 GPUs with roughly 10.1TB of memory in a single standard rack, reducing the cost per million effective tokens by about 40% compared with mainstream international solutions.

Pennsylvania signs AI deceptive advertising law requiring labels on AI-generated ads

Pennsylvania Governor Josh Shapiro signed the AI Deceptive Advertising Law on Thursday, requiring that any advertisement significantly modified or generated by AI—whether image, video, audio, or text—carry a label in the same format as the ad, regardless of where the ad was produced. The law provides exemptions for "parody, satire, commentary, or criticism" and "obviously fictional" content, but key terms such as "significant modification," "obviously fictional," and "creating a false consumer impression" are not clearly defined.

Higgsfield Secures Investment from NVIDIA and TCV, Annualized Revenue Tops $1 Billion

AI video startup Higgsfield announced that NVIDIA and TCV have joined its Series B extension financing, with Jensen Huang calling its founder one of the most talented people in the world. Since its Series B in August, the company's annualized revenue has topped $1 billion, enterprise contracts have grown 71% with month-over-month growth exceeding 100%, and global users have risen from 30 million to more than 40 million. NVIDIA has provided computing power for it all along, from L40 and A40 GPUs to Hopper and Blackwell, and it will subsequently adopt Vera Rubin.

Unitree H2-X humanoid robot side-flip demo fails, mechanical lower leg detaches

During a live demonstration at the "2026 World Physical AI Conference and Unitree Partner Ecosystem Conference," a white humanoid robot lost balance and fell sideways while attempting a side flip landing, and one mechanical lower leg detached from its joint. The robot involved in the accident was the H2-X humanoid robot that Unitree Technology had just released on October 21, officially named "H2-X Kung Fu Legend Guardian." Previously, official promotional videos showed the robot kicking in a car door, lifting a car with its bare hands, and jumping from a platform about 4 meters high to complete a flip, mainly targeting security and emergency rescue scenarios, with a maximum battery life of about 8 hours. As of now, Unitree Technology has not commented on the matter.

Indian workers wear cameras to film work processes, providing data for physical AI training

BBC reports that hundreds to thousands of workers in India are paid to wear head-mounted cameras to record work processes such as haircuts, sewing, and housework; the videos are collected by local companies and sold to tech companies and research institutions for training physical AI that can control real-world robots. Rashdan Technologies pays 5 to 10 dollars for 8 to 10 hours of recording, and Humyn Labs has 96,000 data collectors in Latin America and Southeast Asia. TISS professor Bino Paul believes it is too early to judge whether robots will replace these workers, but calls on the government to regulate AI.

English translation is being prepared.

Sundar Pichai: Google Argon Reaches 25 Trillion Tokens in Usage Within 72 Hours of Launch

Sundar Pichai said that within 72 hours of Google's Argon going live, usage approached 25 trillion tokens, roughly 3 times the usage of Gemini 3.0 over the same period. He also mentioned that the model is currently being served, but compute capacity is constrained. These remarks come from a video on Alex Kantrowitz's YouTube channel.

English translation is being prepared.

English translation is being prepared.

Tenstorrent raises $1.5B at a $6.2B valuation to expand open-weight AI chips

AI chip startup Tenstorrent has completed a $1.5 billion funding round at a valuation of $6.2 billion, more than double its $2.6 billion valuation at the end of 2024. The round was led by Trace Capital and Innovation X, with participation from Cadence Ventures, Eclipse, Fidelity, SanDisk, and T. Rowe Price. CEO Jim Keller said the company decided to remain independent and continue raising funds after demonstrating that its Galaxy servers can run DeepSeek models, and that the new capital will be used to expand chips for open-weight model inference and video generation.