Skip to content

#端侧

Today 1 items
TodayOct 18Sun
Oct 16Fri
  2. TechCrunch · AI51

    Amazon devices chief Panos Panay sees a future where AI devices replace the need to use apps

    Panos Panay, Amazon's SVP of Devices & Alexa, said at TechCrunch Disrupt that personal computing is heading toward ambient computing devices where sensors and AI work proactively in the background — the next generation of users won't open an app for everything but will expect devices to understand their needs, though smartphones will remain important.

  4. The Verge · AI73

    Meta and Apple team up to make Muse AI models run better on Macs

    Meta released Muse Spark Open, an open-weights model it calls the strongest American open-weights model yet, with 816B total parameters, natively multimodal, excelling at long-horizon tasks, complex software engineering and multi-step reasoning. Meta partnered with Apple to optimize the model, which can run on desktop via the latest MLX framework on two Apple M5 Ultra Mac Studios; Apple also published a technical article on MLX, and executives from both companies exchanged public pleasantries on X.

Oct 15Thu
  2. NVIDIA Blog71

    Open From Data Center to Desktop: NVIDIA Expands Nemotron Family With Highest Efficiency Super VL and Bolt Models

    NVIDIA has released the open-source 120-billion-parameter multimodal model Nemotron 3.5 Super VL, plus two on-device models — Nemotron 3.5 Bolt 74B-A9B 24GB (already available) and Bolt E4B 4GB (available October 23) — covering agent deployment from data centers to local hardware.

    Why it matters: NVIDIA discloses parameters, quantization and deployment tiers for the three open Nemotron models, so readers can pick the enterprise, workstation or 4GB on-device option that fits.

  4. Semiconductor Engineering Chips and Packaging52

    Why Open-Weight AI Matters

    Semiconductor Engineering examines how open-weight AI models affect chip architecture, noting the focus is shifting from peak TOPS to memory bandwidth and capacity, sustained throughput, numerical format flexibility, and deployment security.

  5. The Verge · AI49

    My AI doppelganger tells me what to eat

    The Verge senior reviewer Victoria Song spent about 100 days using Twin Health's digital twin service, which combines AI with Dexcom G7 CGM data to predict her blood sugar responses to different foods. The service is not publicly available for download and can only be accessed through employers or private health plans, mainly targeting people with chronic metabolic conditions, and requires continuously uploading body composition, blood pressure and other data to feed the digital twin. She said logging fatigue is a common complaint, and while her behavior changed, her goals of losing weight and reducing medication did not pan out.

Oct 14Wed
  2. Google DeepMind48

    Google DeepMind and Singapore Government collaborate to build fine-tuned AI model for scam detection

    Google DeepMind partnered with Singapore's GovTech to launch an on-device scam detection model, fine-tuned from a lightweight Gemma model on local scam conversation data, that analyzes conversations on-device and warns users of manipulative tactics in real time. Scams account for more than half of police reports in Singapore and have caused over S$4 billion in losses since 2019, with social engineering scams making up as much as 80% of losses. The model uses a two-stage pipeline to preserve battery life, and training data was anonymized and contains no citizens' personal information.

  3. TechCrunch · AI30

    Cerebras CEO says we’re nearing a future where everyone has a personal AI assistant at home

    Cerebras CEO Andrew Feldman said at TechCrunch Disrupt 2026 that the era of at-home AI devices is not far off. He predicted personal AI assistants will run locally, splitting work between devices and the cloud like phones do, and could eventually shop on users' behalf. He cited Meta's Muse and Instinct, a startup valued at $10 billion, and said he already uses a set of agents to manage his schedule and look up who is calling him.

  4. TechCrunch · AI41

    Flox Intelligence is building AI to keep wildlife away from trains, airports, and roads

    Flox Intelligence, a startup incubated at Sweden's KTH, has developed edge devices trained on wildlife behavior that identify animal species locally and play species-specific sounds to drive them away, varying their responses to prevent habituation. About 150 devices are deployed at airports, railways, national parks and other sites across North America, Europe and Australia, with an OEM partnership with Alstom. The company has raised about $5 million, is starting a $10 million Series A, and expects roughly $1.5 million in revenue in 2026.

  5. Hugging Face Blog67

    Run fast, private agents on-device with LFM2.5-24B-A2B

    Liquid AI released LFM2.5-24B-A2B, a reasoning model with 24B total parameters and only 2B activated per token, aimed at fast private on-device agents, published with open weights on Hugging Face.

    Why it matters: The release includes full benchmark comparisons, inference speed and a five-stage training pipeline, so readers can judge whether it fits local agent workflows.

Oct 13Tue
  1. NVIDIA Blog68

    All Systems Go: Global PC App Ecosystem Ready for NVIDIA RTX Spark

    NVIDIA announced that RTX Spark laptops go on sale October 16, with up to 128GB unified memory and 1 petaflop of AI compute, alongside multiple software updates released with ecosystem partners.

    Why it matters: The official roundup covers the software ecosystem around the RTX Spark launch, with specific updates and timing for local agent, creative, and developer tools.

Oct 11Sun
Oct 8Thu
  1. The Verge · AI55

    Google’s AI note-taking app transcribes your meetings completely offline

    Google has released Google AI Edge Foresight, an experimental note-taking app that transcribes meetings and audio files completely offline for free, running on macOS via the on-device EmbeddingGemma 2 model. The app can jot brief points during meetings and generate full notes, and a built-in assistant can answer questions about the content with citations to local files; Google says files, audio, and notes never leave the computer, and it is currently optimized only for Apple Silicon Macs.

  7. NVIDIA Blog72

    NVIDIA, Microsoft Kick Off a New Beginning for Windows PCs With RTX Spark and AI Agents

    NVIDIA and Microsoft announced at the Windows AI and Surface event in San Francisco that they are jointly building agent software and hardware for Windows PCs.

    Why it matters: The article details the specs and launch timeline for RTX Spark and DGX Station for Windows, helping readers assess how realistic running large models locally with resident agents will be.

Oct 7Wed
  3. Google DeepMind68

    EmbeddingGemma 2: an open, lightweight multimodal embedding model

    Google DeepMind released EmbeddingGemma 2, a 740M-parameter model built on the Gemma 4 architecture under the Apache 2.0 license, mapping text, code, images, video and audio into a shared embedding space.

    Why it matters: The official release provides concrete metrics such as parameter count, memory footprint and benchmark scores, helping readers judge whether it suits on-device retrieval scenarios.

Oct 2Fri
Sep 24Thu
  1. Hugging Face Blog60

    Accelerating vision-language models with LFM2.5-VL-DSpark

    Liquid AI has released experimental DSpark draft models for the vision-language model LFM2.5-VL-3B, using speculative decoding to trade for faster inference without changing output quality — decoding speeds up to 3.13x on-device and 2.66x on H100, with end-to-end gains up to 2.62x and 2.27x respectively.

    Why it matters: The article details a speculative decoding acceleration scheme for vision-language models with concrete speedups and memory overhead, plus integration guidance for llama.cpp, MLX-VLM, and SGLang.

Sep 22Tue
  1. Hugging Face Blog70

    Transformers now runs llama.cpp quants

    Hugging Face has announced that transformers can load GGUF models — passing a gguf_file to from_pretrained lets you run quantized checkpoints from the Hub on your local machine.

    Why it matters: Hugging Face provides official usage of GGUF in transformers plus benchmarks against llama.cpp, helping readers pick a viable local inference path.

  2. Hugging Face Blog48

    Jun Kim, oMLX creator and maintainer, joins Hugging Face to support the MLX community

    Jun Kim, creator and maintainer of oMLX, has joined Hugging Face. oMLX will shift from a side project to one with full-time maintenance and funding, while remaining Apache 2.0 open source and led by Jun. One team focus is simplifying the pipeline from transformers model definitions to reference MLX implementations, so new transformers models run on MLX faster.

Sep 14Mon
  1. TrendForce Insights Semiconductor Industry Insights55

    Why 4hi HBM May Not Be the One-Size-Fits-All Fix

    TrendForce says 4hi HBM samples will enter supplier views in early 3Q26, with NVIDIA and other AI chip vendors evaluating low-stack configurations to cope with the HBM shortage. The spec has real use cases in edge AI scenarios like autonomous driving and robotics; SK hynix previously supplied HBM2E to Waymo in 2024 with automotive certification, but near-term demand contribution remains very low.

Aug 25Tue
Aug 21Fri
  1. Hugging Face Blog66

    Up to 3.2x Faster Inference with LFM2.5-DSpark

    LiquidAI released DSpark draft model checkpoints for the LFM2.5 series' LFM2.5-1.2B-Instruct, LFM2.5-2.6B, and LFM2.5-8B-A1B, speeding up inference via speculative decoding without changing output quality; throughput on H100 improves up to 3.18x, and up to 2.87x on edge devices.

    Why it matters: The official release provides measured speedup data for the DSpark speculative decoding draft models along with llama.cpp and SGLang usage, so edge deployers can compare and judge the benefits.

Jun 27Sat