Gate News message, April 28 — NVIDIA announced the launch of Nemotron 3 Nano Omni, an open-source multimodal inference model designed to provide a unified foundation for enterprise-grade AI agents. The model combines vision, audio, and language capabilities to enable AI systems to achieve up to 9x e

Gate News message, April 28 — NVIDIA has released Nemotron 3 Nano Omni, an open-source multimodal model featuring a 30B-A3B mixture-of-experts (MoE) architecture with 256K context window support. The model unifies processing of video, audio, image, and text inputs in a single framework. Compared

Gate News message, April 23 — Anker Innovation, a leading global consumer electronics brand, unveiled its first neural network Compute-in-Memory (CIM) AI audio chip called Thus™ on April 22. CEO and founder Yang Meng revealed the company's three major technology initiatives in edge AI: the Thus™ chi

Gate News update, April 20. According to Gate market data, as of the time of publication, ORDI (Ordinals) is trading at $4.76. It is up 10.13% over the past 24 hours, with a high of $5.36 and a low of $4.05. The 24-hour trading volume is $8.4388 million. The current market cap is about $100 million. The Ordinals protocol enables the inscription of NFTs and tokens on Bitcoin by writing content into the smallest unit of Bitcoin, the satoshi. Bitcoin’s total supply of 21 million coins is further divided into 21 trillion satoshis. Users can permanently record content such as text, images, audio, and video on satoshis through the Ordinals protocol, without relying on separate tokens, sidechains, or modifying Bitcoin itself. OR

Gate News update: On April 16, according to Gate market data, as of the time of publication, OORT (Oort) is trading at $0.01195. Over the past 24 hours, it is up 49.67%, with a high of $0.015 and a low of $0.00792. Its current market cap is about $8.8926 million. OORT is a decentralized AI infrastructure that protects data privacy and saves costs. By integrating global computing and storage resources, OORT can support trusted AI applications. OORT DataHub is a decentralized cloud platform where users can collect and preprocess data, including images, audio, or video, to improve AI and machine learning models. Users can earn OORT tokens by completing micro-tasks, verifying data, and inviting friends, among other ways.

Gate News message. On April 16, according to Gate market data, as of the time of writing, ORDI (ORDI) is trading at $4.46. It is up 72.87% over the past 24 hours, with a high of $4.74 and a low falling to $2.54. The 24-hour trading volume is $3.5059 million. The current market cap is approximately $93.765 million. ORDI is an inscription asset based on the Ordinals protocol. The total supply of Bitcoin is 21 million coins. Each coin can be subdivided into 100 million satoshis (the smallest unit of Bitcoin), for a cumulative total of 21 quadrillion satoshis. Inscriptions create by writing content into satoshis using the Ordinals protocol, without needing separate tokens, sidechains, or modifying Bitcoin. The protocol writes information to each satoshi, including text, images, audio, and video, etc.

Xiaohongshu's AI team has open-sourced Relax, a reinforcement learning engine for multimodal training that efficiently processes image, audio, and video inputs through advanced parallelism and asynchronous pipelining.

Gate News message, April 12, Musk’s social app XChat will officially launch on the App Store on April 17. Musk previously said that XChat’s goal is to replace the old Twitter direct message system and build a fully encrypted platform supporting text, file transfers, and audio and video calling.

Google released the multimodal model Gemma 4 on April 3. It supports text, image, and audio inputs and generates text outputs. Its context window can hold 256,000 tokens, supports 140 languages, and is designed for a variety of environments, including four different model sizes.

Alibaba Qwen3.5-Omni, a fully multimodal large model, was released on March 30. It includes Instruct versions in three different sizes, supports long context, and supports multiple audio and video inputs. The model is pretrained on massive amounts of data, significantly improving multilingual capabilities, and supports 113 speech recognition languages and 36 speech generation modes.