Search “Speech”

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.

How to get Google Assistant back on Android and replace Gemini

Some Android users can choose the old assistant again after Gemini has fully taken over by uninstalling updates to the Google app and then reinstalling Google Assistant from the Play Store. This procedure also requires turning off automatic updates for the Google app, but it means missing security updates, and the restored Assistant does not respond to wake words such as "Hey Google" and can only be activated with the physical button. The article considers this very likely a system loophole that Google will close sooner or later.

Why DUV Controls Hold Up: Saif Khan on the U.S.-China AI Chip Rivalry

Saif Khan, a former U.S. National Security Council and Commerce Department official, predicts in a new DUV report from the new think tank CTS that if exports of the relevant equipment stopped now, by the mid-2030s China could produce roughly 100 times fewer AI chips than the United States, making DUV controls the most critical lever for the United States to shape the AI balance of power. In a ChinaTalk interview, he also discussed how robotics will become the next frontier in U.S.-China AI competition, and how AI is changing policy research itself.

Bluetti Balco 260 balcony power station storage tested: supports AC charging

In testing, the Bluetti Balco 260 balcony power station storage supports AC charging. The main unit has a built-in 2.56 kWh LiFePO4 battery; with the BC260 expansion battery, total capacity is 5.12 kWh, and up to five expansion batteries can be connected for 15.36 kWh. It features four independent MPPT inputs that together accept up to 2400 W of solar, and it supports AC surplus charging via the Shelly Pro 3EM, with prices starting at 649 euros. In testing, discharge efficiency at 800 W was good, but at a constant 200 W discharge the usable AC energy was slightly lower, and standby loss was about 20 W.

Microsoft Surface Laptop Ultra 128GB Review: Powered by RTX Spark N1X

The Microsoft Surface Laptop Ultra is one of the first Windows laptops to feature the NVIDIA RTX Spark N1X (the consumer brand of the GB10 SoC), offered in 18-core and 20-core versions, with up to 128GB of LPDDR5X memory and a 1TB PCIe Gen5 SSD. This 15-inch model weighs 2.0kg and is 19.16mm thick, and features a 3270×2180 PixelSense touchscreen with support for a 120Hz variable refresh rate and up to 2000 nits of brightness, positioned to compete with Apple's MacBook Pro.

Claude web and desktop now offer a Chinese interface, but mainland China is still not on the list of supported regions

Claude web and desktop have added a language selector, and the interface now supports Chinese (Simplified) and Chinese (Traditional), among more than thirty languages; the mobile app has not yet followed suit. In the billing country drop-down menu on the checkout page, China appears, and both the $20 Pro and $200 Max plans are selectable, but the official list of supported regions still does not include mainland China, Hong Kong, or Macao. Anthropic has previously imposed account bans and identity verification restrictions on access from unsupported regions; this Chinese interface is mainly aimed at overseas Chinese users.

Microsoft Surface Laptop Ultra review: first to feature NVIDIA RTX Spark

The Microsoft Surface Laptop Ultra is the first Windows laptop to feature the NVIDIA RTX Spark SoC (the consumer brand for GB10), available with an 18-core or 20-core N1X processor, equipped with 128GB of LPDDR5X memory, a 1TB PCIe Gen5 SSD, and a 92Wh battery. It features a 15-inch 3270×2180 PixelSense touchscreen with up to 2000 nits of brightness and support for a 120Hz variable refresh rate; the device weighs 2.0kg and is 19.16mm thick, and is regarded as Microsoft's high-end product positioned against the MacBook Pro.

English translation is being prepared.

ElevenLabs Renews Multiple Security Certifications and Adds FedRAMP

ElevenLabs announced the renewal of certifications including ISO/IEC 27001:2022, ISO 27701:2019, ISO/IEC 42001:2023, ISO 27017/27018, SOC 2 Type II, PCI DSS Level 1, and Cyber Essentials Plus, and added SOC 2 Type I and FedRAMP 20x Class A. FedRAMP is a rigorous U.S. government certification that assesses whether cloud services can handle federal data, and Class A is a new entry path under FedRAMP 20x.

English translation is being prepared.

English translation is being prepared.

NVIDIA Expands Nemotron Model Family with New Super VL and Two Bolt Models

NVIDIA has added three models to the Nemotron open-source model family, covering enterprise workflows, local AI agents, and conversational applications across data centers, desktops, and edge devices. These include the 120-billion-parameter Nemotron 3.5 Super VL, which NVIDIA claims delivers up to 5x the throughput of the next-best comparable model, with accuracy approaching that of its largest reasoning model at roughly one-fifth the parameter count; there are also two small models in the Nemotron 3.5 Bolt series: 74B-A9B 24GB for everyday agent tasks on RTX and DGX Spark, and E4B 4GB for conversational and physical AI agents. Super VL and Bolt 74B-A9B are already available, and Bolt E4B is scheduled for release on October 23.

The Ministry of Commerce and the Ministry of Science and Technology plan to revise the technology export catalog, adding 23 artificial intelligence technologies

On October 16, the Ministry of Commerce and the Ministry of Science and Technology issued an announcement proposing to revise the Catalog of Technologies Prohibited or Restricted from Export in China, soliciting public comments on 40 technologies in three fields: artificial intelligence, biomedicine, and photovoltaics, while simultaneously implementing pre-export registration management. The artificial intelligence field involves 23 technologies related to large models, agents, computing power clusters, and information processing; the restricted export section lists specific indicators such as AI chip design, high-speed interconnection of computing power clusters, and fault tolerance in distributed training. The announcement specifies that technologies that have entered the public domain through open source, published academic papers, and other means are not included in the scope of management, and that enterprises using restricted large models to provide external API calls, APP or web interaction, SaaS applications, and local on-device lightweight deployment do not fall within the scope of large model technology export management.

Hugging Face Releases Beam: A 501B Open Source Model Supporting 1M Context and Self-Hosting

Hugging Face has released Beam, a 501B-parameter sparse MoE open source model aimed at long-text reasoning and agent tasks with 1M context, activating only about 23B parameters per token. The model uses 52 layers of hybrid attention, with 43 layers using a 512-token sliding window and 9 layers using global attention; the KV cache grows by 36 KiB per token, about one-sixth that of a full-attention approach. The official release provides three checkpoints—BF16, FP8, and NVFP4; NVFP4 can compress the weights from about 1 TB to about 280 GB, and day-0 support is already available in transformers, vLLM, and SGLang.

English translation is being prepared.

English translation is being prepared.