Search AI Products and News

The Kimi Open Platform launched the Kimi K2 High-Speed Version (kimi-k2-turbo-preview), increasing the output speed from 10 Tokens per second to 40 Tokens per second while maintaining the original parameter scale, significantly improving response efficiency. This upgrade is mainly targeted at applications requiring higher real-time performance, aiming to optimize user experience.

kimi-k2-turbo-preview, the high-speed version of Kimi K2, is officially released. According to the information, this high-speed version model maintains the same parameter settings as the original kimi-k2, but achieves significant performance improvements. The output speed has jumped from 10 Tokens per second to 40 Tokens per second, providing users with a smoother and more efficient experience.

San Francisco AI startup Deep Cogito launches the Cogito v2 series of open-source large models, focusing on machine intuition and self-improvement capabilities. The series includes four dense and mixture-of-experts models with 70B to 671B parameters. The flagship 671B MoE model has a reasoning efficiency that is 60% higher than DeepSeek R1, and its performance is close to Claude4Opus. Its core technology involves internalizing reasoning paths into model weights through introspective reasoning, enabling continuous evolution while in use. Tests show that the model can

Meta CEO Zuckerberg published an open letter titled 'Personal Superintelligence,' announcing the company's commitment to developing personal superintelligent systems. He proposed a decentralized approach that differs from the industry's centralized perspective, advocating for superintelligence empowering each individual. Zuckerberg remains optimistic about technological progress, believing it will usher in a new era of individual empowerment, but he also emphasized the need for careful management of security risks. Notably, the letter hints at potential future adjustments to the open-source strategy, shifting towards a more conservative proprietary model in core technology areas.

Meta upgrades the Avatar system, adding more than 10 preset body types and detailed area adjustment features, supporting 18 postures and over 50表情 stickers. The AI styling assistant can quickly generate virtual avatars based on text descriptions, currently available only in the US and Canada. Parametric facial control enables billions of combination possibilities, allowing users to precisely customize their virtual avatars.

Meta CEO Zuckerberg offered $1B to 12 TML AI talents, but none accepted. Meta denies some reports. Zuckerberg contacted candidates via WhatsApp, promising a top AI assistant and open-source strategy. However, Meta's vague roadmap and TML's recent $12B funding deterred talent, who prioritized growth over salary.....

Apple loses four AI experts to Meta in a month, including key researcher Zhang Bowen and former AI head Pang Ruoming (compensation over $200M). Meta's aggressive hiring for its super AI team outpaces Apple's retention efforts, highlighting Apple's AI talent struggle and strategic concerns.....

Meta plans to launch a new smartwatch in Sept 2025, made by Huaqin, featuring a camera for use with Meta AI glasses. Development started in 2022 but faced delays. Meta is also developing a neural smartband with Carnegie Mellon, enhancing interaction and aiding disabilities, building an AI wearable ecosystem.....

According to EssilorLuxottica's financial report, sales of the Ray-Ban Meta smart glasses, developed in collaboration with Meta, increased by more than double, driving the company's total sales to 14.02 billion euros (+7.3%) in the first half of the year. The company emphasized its leading position in the smart glasses sector, believing that glasses will become the next-generation computing platform integrating AI and sensing technologies. Recently, they launched the Oakley Meta product and renewed a ten-year partnership agreement. They may also launch Prada smart glasses in the future. Brands such as Ray-Ban have helped the company establish a strong position in the smart wearable market.

Pudong New Area launched a seed fund with a total scale of 2 billion yuan for artificial intelligence. The first phase is 500 million yuan, making it the first state-owned capital venture capital fund in China focusing on the pre-seed stage. It has signed 18 projects. The AI industry scale in Pudong exceeds 160 billion yuan, accounting for 40% of the city, and has gathered more than 200 specialized model enterprises. In the future, five major measures will be implemented: releasing over 30 application scenarios, building a heterogeneous robot training field, establishing a full-cycle fund matrix, optimizing industrial policies, and providing talent services. The goal is to add 1,000 AI companies within three years, and the core output value will exceed 200 billion yuan, aiming to build a comprehensive AI ecosystem.

VLM2Vec-V2, a multimodal embedding framework developed by Salesforce Research and other institutions, breaks through the limitations of traditional models. Based on the Qwen2-VL architecture, this framework innovatively unifies image, video, and visual document retrieval tasks. It adds five new evaluation tasks to expand the MMEB dataset. With key technologies such as dynamic resolution and M-RoPE, it achieved an average score of 58.0 across 78 datasets, leading in performance, especially in video tasks. Although its document retrieval is slightly behind ColPali, it

Zhiyuan Robotics announced the open-source plan for the Zhiyuan Lingqu OS, an embodied intelligence operating system, at the 2025 World Artificial Intelligence Conference. The humanoid robot Lingxi X2 demonstrated intelligent interaction capabilities through natural dialogue. The system adopts a layered open-source model, including underlying middleware, agent framework, and toolchain, with plans to launch in the fourth quarter. Zhiyuan Robotics has already achieved application in fields such as industrial logistics, and its Qiyuan general embodied large model won the SAIL Star Award. The company will continue to improve the operating system and promote the development of embodied intelligence industry.

The Meta research team introduced an innovative architecture called AU-Net, breaking through the limitations of traditional tokenization techniques. The model learns directly from bytes, dynamically combining bytes into hierarchical sequence representations through a self-regressive U-Net structure. Its contracting path compresses bytes into semantic units, while the expanding path restores information and fuses details, with skip connections preserving local features. A multi-stage pooling strategy is used to extract semantics at the word and phrase levels, and multi-linear upsampling optimizes information fusion. The self-regressive generation mechanism ensures text coherence and significantly improves inference efficiency. This architecture

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

Meta established a Super Intelligence Lab and has recruited 44 top AI talents, with 40% coming from OpenAI and 50% being Chinese. Zuckerberg is shifting his strategic focus to AI, willing to offer high salaries to attract talent, with a maximum signing bonus of $200 million. The new team has 75% with doctorates and 70% as researchers, with annual salaries ranging from $10 million to $100 million. This move highlights Meta's determination to secure leadership in AI and reflects the intensifying competition for talent in the industry.

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.

Meta forms 'Super AI Lab' with 44 top talents (50% Chinese, 40% ex-OpenAI), shifting focus from metaverse to AI. Offering huge incentives like $200M signing bonus. Team: 75% PhDs, 70% researchers.....

Meta reorganized its AI structure to establish a Superintelligent Lab, integrating 3400 employees, led by Alexandr Wang as Chief AI Officer. The new structure is divided into four teams: AGI Basic Research, AI Product Development (including Meta AI Assistant), the Basic AI Lab led by Yann LeCun, and a group focused on Llama5 development. Meta is offering high salaries to attract talent from companies like OpenAI and Apple, but this has raised doubts within the original team about the influx of high-paid outsiders. Recently, two AI leads from Apple have joined.