Search AI Products and News

In the 2025 RoboCup Humanoid League, the Tsinghua Huoshen team won their first championship by using the HumanoidKick algorithm developed jointly by ByteDance Seed and Tsinghua University. The algorithm enables robots to complete the entire process from finding the ball to kicking it through visual deep reinforcement learning, allowing them to respond in real-time to changes on the field. The competition, established in 1997, is the top event in the field of robot technology. The Tsinghua Huoshen team won convincingly in both the group stage and elimination rounds, ultimately winning 5:2. This breakthrough marks a significant milestone for China.

Shenzhen Leju Robotics has secured a new round of funding, with its registered capital increased to 2.27 million yuan, and investors including Dongfang Jinggong. The company, established in 2016, has accumulated over tens of millions in financing. It has transitioned from educational robots to the industrial field. Its Aelos educational robot has been adopted by more than 4,000 schools. In 2023, it launched a humanoid robot based on open-source HarmonyOS, and collaborated with Huawei Cloud and China Mobile to develop 5G-A robots, enhancing application capabilities in industrial scenarios. The company plans to deliver 1,000 humanoid robots by 2025, addressing labor shortages in manufacturing and promoting intelligent production.

Chinese robot manufacturer Unitree has launched the world's first full-size humanoid robot R1 priced below $6,000, at only 39,900 yuan. This 25-kilogram robot features 26 joints and can perform high-difficulty movements such as backflips, equipped with a multi-modal AI system. Compared to similar products, the R1 is only one-third the price of Tesla's Optimus, thanks to China's supply chain advantages and lightweight design. Unitree's move comes as it prepares for an IPO, potentially triggering a price war in the industry. Although the R1 is not yet suitable for home use,

The unitree robot G1 and the humanoid robot DeREK won the fighting competition, but suddenly lost control. Due to being in mid-air, it accidentally entered walking mode, causing packages to fly around. CEO Cix explained that the emergency brake takes 5 seconds to take effect, during which damage occurred. The Ethernet cable eventually came loose, stopping the incident. A similar issue had occurred in May, and netizens questioned the safety design, suggesting that multiple safety mechanisms should be improved rather than rushing to raise funds. The incident has sparked extensive discussions about robot safety.

Musk opens Tesla Diner in LA, blending fast food, cinema & charging. Features 80 chargers & robot waiters, earns $47K in 6 hours. Plans Texas expansion to showcase tech.....

Brazilian programmer won $50K in K Prize AI contest with only 7.5% accuracy, revealing AI model limits. $1M prize offered for 90% accuracy in GitHub-based test.....

The Kimi team released the Kimi K2 technical report. This trillion-parameter large model, with 32 billion activated parameters, won the global open-source model competition, with performance comparable to top closed-source models such as GPT4.5. Its innovations include: using the MuonClip optimizer to process 15.5 trillion token data without loss; developing a multi-domain tool data synthesis pipeline; pioneering the restatement method to improve training efficiency, achieving the effect of ten rounds of original data in one round of rewritten data; optimizing model performance through supervised fine-tuning and reinforcement learning. The model is based on N

AI code review startup Greptile is currently raising $30 million in Series A funding, with an estimated valuation of $180 million. Founded by 22-year-old Dasksh Gupta, the company secured a $4 million seed round after completing Y Combinator's incubation program. Its core product is an AI tool that can review code like an experienced engineer. However, the sector is highly competitive, with main competitors Graphite and Coderabbit having raised $52 million and $16 million respectively.

Recently, ByteDance's Seed team officially launched a new Vision-Language-Action Model (VLA) called GR-3. This model demonstrates breakthrough capabilities in the field of robotic manipulation, not only understanding language instructions that include abstract concepts, but also precisely handling flexible objects. It also has the ability to generalize quickly to new tasks and recognize new objects. This achievement is seen as an important advancement toward a 'general-purpose robot brain'. Traditional robotic manipulation models often rely on large amounts of robotic trajectory data for training.

DeepSeek's Q2 downloads plunged 72.2% to 22.58M, losing market appeal. Doubao emerged as new leader while Kimi declined. Experts cite lost cost edge, cheap APIs, user churn, weak use cases, and R2 delays. Meanwhile, 'AI+Office' grew 190.2% and 'AI+Education' hit 10M users, signaling intense competition demanding constant innovation.....

AI giants OpenAI and Google DeepMind have engaged in a fierce competition over their IMO gold medal achievements. Both companies announced that their AI models correctly solved 5 out of 6 problems in the 2025 International Mathematical Olympiad, surpassing most human competitors. This breakthrough demonstrates progress in AI's ability to handle fuzzy reasoning, but it has sparked a debate over who announced first: OpenAI released the news on Saturday, which was criticized by Google for not following the official process. Google stated that their results were recognized by the official IMO, while OpenAI hired a former IMO medalist to evaluate the results. This debate

Zhuji Power has received strategic investment from JD.com, accelerating the mass production of humanoid robots and the research and development of embodied intelligence technology. The company has launched the IDS Collaborative Innovation Strategy, serving innovators, developers, and system integrators by providing an open technology platform. A full-size humanoid robot is expected to be launched in 2025, with the ability to adapt to complex environments. The bipedal robot TRON1 has already been applied in areas such as inspection and security, and a developer kit has been released. The company's LimX VGM large model enables cross-platform operation capabilities, and more visual-language-action large model results will be released this year.

In 2025, open-source large models will show three major trends: 1) MoE architecture becomes mainstream, with DeepSeek-V3 (67.1 billion parameters) and Qwen3-235B (235 billion parameters) each having unique designs in their expert systems; 2) small models break through performance bottlenecks, with SmolLM3-3B adopting position-encoding-free technology, and Qwen3-4B achieving lightweight efficiency; 3) models show significant differentiation, with Llama3.2 focusing on general tasks, while Kimi-K2 (1 trillion parameters) excels in complex reasoning.

Chinese humanoid robot company Yushu Technology has initiated A-share listing tutoring, with CITIC Securities as the tutoring institution. The company became popular after appearing in the Spring Festival Gala, and recently won the WIPO Global Award, becoming the only Chinese company to win the award. It has completed share reform and increased its capital to 364 million yuan, having received investments from well-known enterprises such as Geely and Tencent. A-shares companies that indirectly participate in Yushu Technology have shown strong stock performance, with an average annual increase of over 18%, indicating market confidence in the humanoid robot industry.

Apple's MLX framework now supports CUDA, marking a strategic shift to embrace NVIDIA's ecosystem. This enables developers to train models on NVIDIA GPUs and deploy on Apple devices, enhancing efficiency and bridging platform gaps.....

Meta accelerates AI infrastructure, targeting a 1GW 'Prometheus' supercomputer with 1.3M NVIDIA H100 GPUs (2 exaflops) by 2026, plus 5GW 'Hyperion' cluster. Plans $60-65B investment by 2025 for AI/data centers, competing with OpenAI/xAI. Commits to open-source and privacy despite environmental concerns.....

A dramatic acquisition has recently taken place in the AI coding field: Cognition acquired Windsurf company. Previously, this company had experienced a $2.4 billion reverse talent acquisition by Google and an unsuccessful $3 billion acquisition offer from OpenAI. Windsurf generates $82 million in annual revenue, has 350 enterprise clients, and tens of thousands of daily active users. After the acquisition, Cognition will integrate Windsurf's AI development environment with its own Devin coding assistant and regain access to the Claude AI model. This deal marks another significant move in the competition.

HuggingFace launched the open-source desktop robot Reachy Mini, triggering a rush to buy. This 28-centimeter-tall robot, starting at 299 dollars, achieved sales of over 130,000 euros in five hours. It features a six-degree-of-freedom head, wide-angle camera, and modular design, supporting 15 preset actions and community behavior sharing. It is both a toy and an AI learning platform. As an AI platform valued at 4.5 billion dollars, HuggingFace has acquired the robotics company PollenRobotics and launched the LeRobot project.

Tesla's first themed restaurant opens in LA on July 21, featuring Optimus robot waiters, UFO-style neon-lit architecture, 200 seats, rooftop bar, 13.7m LED screen, 80 V4 superchargers, and tech-inspired menu with 'autopilot delivery'. Combines dining, charging, and robotic service.....

AWS upgraded SageMaker with model observability and local IDE integration. HyperPod now monitors training performance and connects local dev environments to cloud. GPU cluster management was optimized for flexible resource allocation.....