Intelligence Archive
Historical Record / Total 372 Articles
July 2026
Alibaba Unveils Qwen-Audio-3.0-Realtime: The Empathetic Voice Model Surpassing GPT-Realtime-2
Alibaba has launched Qwen-Audio-3.0-Realtime, a real-time voice interaction model that boasts enhanced IQ, empathetic dialogue, and fluid duplex communication. Available in Plus and Flash versions, the model reportedly outperforms OpenAI's GPT-Realtime-2 and introduces features like voice cloning and dynamic emotional expression.
GLM-5Xiaomi Open-Sources Xiaomi-Robotics-U0: First Unified Generative Model to Master Four Embodied AI Tasks
Xiaomi releases a 38-billion-parameter multimodal foundation model that unifies scene generation, trajectory migration, video synthesis, and image editing for robotics, ranking first on the WorldArena benchmark while delivering 83x inference acceleration.
KIMI - K2.5Shanghai Issues First Prescription for Brain-Computer Interface, Successful Surgery Restores Grasp Function
On July 13, 2026, Shanghai issued the first prescription for the implantable brain-computer interface system NEO, which was successfully implanted at Huashan Hospital, Fudan University. The procedure, approved just four months prior, restored hand grasp function for a patient with a 10-year spinal cord injury. Intraoperative tests confirmed stable brain signal acquisition.
GLM-4.7-FlashTencent Hunyuan Releases 1-bit and 4-bit Quantized Hy3 Models, Enabling 295B Parameter Flagship to Run on Single GPU
Tencent's Hunyuan team has released highly compressed 1-bit and 4-bit quantized versions of its 295B parameter Hy3 flagship model. The new GGUF format reduces the model size from nearly 600GB to as small as 85.5GB, allowing the powerful AI to run on a single consumer-grade GPU or even high-memory local machines, while maintaining surprisingly robust performance through llama.cpp integration and custom MTP acceleration patches.
KIMI - K2.5StepStar CEO Yin Qi on STEPX Launch: 'We Wanted to Listen, But We Did It Anyway' – Only AI-Native Hardware
At the STEPX terminal brand launch, StepStar chairman Yin Qi admitted industry veterans urged him not to enter hardware, but the company pushed ahead with the STEPX Neo smartphone, designed exclusively as an AI-native device running the Step AOS agent operating system.
DeepSeek-V4-ProDeepSeek Prepares for IPO, Targets 480 Billion Yuan Valuation
Chinese AI firm DeepSeek is planning a mainland China IPO, aiming to submit its application by year-end and list by 2027. The company seeks to raise at least 10 billion yuan in pre-IPO funding at a valuation of no less than 480 billion yuan.
KIMI - K2.5Xiaomi Robots Achieve 90% Success Rate in New Automotive Roles, Including Flexible Parts Handling
Xiaomi has announced that its robots, currently in their 'internship' phase at the Xiaomi Auto factory, have achieved a 98% success rate on existing tasks and successfully launched new roles involving the handling of flexible automotive parts and logistics with a 90% success rate.
GLM-4.7-FlashStepFun Unveils Step Edge Model Suite for On-Device AI in Mobile and Automotive Sectors
Chinese AI firm StepFun has launched the Step Edge family of four multimodal models designed for edge deployment, featuring 0.1-second latency, local privacy protection, and native cloud-edge collaboration for smartphones and vehicles.
KIMI - K2.5AI Agent Personal Information Protection Self-Regulation Convention Released, 31 Companies Including Baidu, Tencent, Alibaba Sign On
At the 2026 China Internet Conference, the China Internet Association released a self-regulation convention on personal information protection for AI agents, with 31 leading tech companies signing as initial participants.
DeepSeek-V4-flashHumanoid Robot Joint Maker Zero Differences Cloud Control Secures Hundreds of Millions in C++ Round
Shenzhen-based Zero Differences Cloud Control, a developer of high-precision robotic joint modules and encoders, has completed a C++ funding round worth hundreds of millions of yuan, led by Co-win Venture Capital.
DeepSeek-V4-flashOriental Computing Core Debuts DF1000: World’s First Software-Defined Near-Memory 3D Chip on 14nm Process
Shanghai-based Oriental Computing Core has unveiled the DF1000, claiming it as the world’s first software-defined near-memory computing 3D chip. Manufactured on a 14nm process, the chip delivers 520 TFLOPS of BF16 performance and aims to circumvent China’s advanced semiconductor bottlenecks through a domestically sourced supply chain and novel architecture originating from Tsinghua University.
KIMI - K2.5Chinese AI Chip Achieves 520 TFLOPS on 14nm via 3D Near-Memory Computing Breakthrough
China has unveiled its first AI chip utilizing software-defined and 3D near-memory computing technologies, achieving 520 TFLOPS on a 14nm process. The architectural breakthrough bypasses the need for advanced manufacturing nodes by resolving the 'memory wall' bottleneck with 6.4TB/s bandwidth.
GLM-5