Shadow-Free Augmented Reality Makes Illusions More Realistic

GLM-5.2: Built for Long-Horizon Tasks

Z.ai vừa ra mắt GLM-5.2, mô hình nguồn mở 753 tỷ tham số (MIT license) tối ưu cho các tác vụ lập trình dài hạn nhờ nhiều cải tiến như cửa sổ ngữ cảnh 1 triệu token, kiểm soát "effort-level" cân bằng hiệu suất-latency, và kiến trúc IndexShare giúp giảm 2,9 lần FLOPs/token. Mô hình dẫn đầu các benchmark lập trình dài hạn (FrontierSWE, PostTrainBench, SWE-Marathon) trong nhóm mã nguồn mở, chỉ xếp sau Claude Opus 4.8, đồng thời hỗ trợ các framework suy luận phổ biến như vLLM và SGLang.

Lập trình viên nên đọc bài này vì GLM-5.2 là một mô hình AI mạnh mẽ cho các nhiệm vụ lập trình dài hạn, giúp tối ưu hóa hiệu suất và độ chính xác trong việc xử lý các dự án phức tạp, từ việc viết code đến tối ưu hóa logic, với các tính năng như hỗ trợ context rộng và kiến trúc hiệu suất cao.

Privacy-Aware Infrastructure in the AI-Native Era: An Asset Classification Case Study

Meta's engineering team describes a hybrid pattern for privacy-aware infrastructure (PAI) asset classification at scale. The core approach combines LLMs for handling ambiguous or novel data assets with deterministic, versioned rules for routine enforcement. Key principles include: building structured 'evidence briefs' from distributed context (code lineage, ownership, semantic annotations) rather than raw fields; keeping human-reviewed labels separate from model-generated recommendations; using a multi-panel LLM judge with Cohen's kappa for quality control; and progressively distilling stable LLM-discovered patterns into auditable deterministic rules. In production, ~85% of traffic is resolved by fast deterministic rules, with LLM fallback for the remaining ~15%. The system includes self-regulation mechanisms (tuning controller states) to prevent runaway optimization loops. Lessons learned emphasize that context quality beats prompt quality, accuracy alone is insufficient for imbalanced taxonomies, and distillation into deterministic rules is the sustainable production model.

Firefly AIBOX-9075 Edge AI box features Qualcomm IQ-9075 SoC with 200 TOPS NPU, 36GB LPDDR5

Firefly has launched the AIBOX-9075, an industrial Edge AI box powered by the Qualcomm DragonWing IQ-9075 SoC. It delivers up to 200 TOPS of AI performance via a Hexagon Tensor Processor NPU, paired with 36GB LPDDR5 ECC memory and 128GB UFS 2.2 storage. The device targets edge AI workloads including private LLM deployment, robotics, and computer vision. It supports popular AI frameworks (TensorFlow, PyTorch, ONNX) and LLMs such as DeepSeek-R1, Llama, and Gemma, achieving up to 22 tokens/s with Llama2-7B. Connectivity includes dual 2.5GbE with TSN, optional Wi-Fi 6, 4G/5G, and 8x GMSL2 camera inputs. The rugged aluminum enclosure operates from -40°C to 85°C. It runs Ubuntu and Yocto Linux and is priced at $1,239 from the Firefly store.

$449 CHUWI UniBook 14-inch laptop is powered by an Intel Core 3 304 Wildcat Lake processor

CHUWI has opened pre-orders for the UniBook, a 14-inch laptop priced at $449 (regular $499) powered by Intel's new Core 3 304 'Wildcat Lake' processor. The penta-core SoC is built on Intel 18A and features an Xe3 GPU and 15 TOPS NPU. The laptop ships with 8GB LPDDR5 RAM, a 256GB NVMe SSD, a 1920×1200 IPS display, Wi-Fi 6, and a 53.38Wh battery rated for up to 13 hours. Early PassMark benchmarks suggest performance comparable to Apple's A18 Pro. Shipping is expected to start July 20, 2026, making this one of the first consumer Wildcat Lake laptops available for purchase.

This ASUS Vivobook S16 with an AMD Ryzen AI 7 CPU hits its lowest price ever, but you'll need to act fast before Prime Day ends

The ASUS Vivobook S16 is on sale during Amazon Prime Day 2026 at its lowest price ever — $880 (down from $1,095). It features an AMD Ryzen AI 7 350 CPU, a 16-inch 3K OLED display at 120Hz, 16GB RAM, 1TB storage, and a 75Wh battery rated for up to 14 hours. The laptop is thin and light at 0.55 inches and 3.31 lbs, with integrated graphics rather than a discrete GPU. It includes a solid port selection: HDMI, two USB-C, microSD, audio jack, and two USB-A ports.

TPUs vs GPUs: When to Choose What for AI/ML Workloads

A practical framework for choosing between TPUs and GPUs for AI/ML workloads, covering silicon architecture differences, use-case fit, and total cost of ownership. TPUs excel at large-scale JAX-based pretraining (100B+ params) on GCP with committed-use discounts, but their static shape requirements, GCP-only availability, and smaller ecosystem make GPUs the default for most teams. GPUs dominate due to PyTorch/CUDA ecosystem maturity, dynamic shape support, multi-cloud portability, and viable spot automation. The post also covers GPU cost optimization strategies including rightsizing via DCGM, spot instance automation, MIG partitioning, and inference density improvements, with Cast AI promoted as a solution for automating these optimizations.

Discounted 1.25 GHz Raspberry Pi 4 Model B shows up online

Cytron bán Raspberry Pi 4 Model B "Special Value Board" với RAM 4GB/8GB, xung nhịp 1.25 GHz (thấp hơn chuẩn 1.8 GHz) do lỗi sản xuất nhưng vẫn hoạt động ổn định. Giá giảm còn $87.25 (4GB) và $147 (8GB), tiết kiệm $13–$18, nhưng chỉ còn 258/189 sản phẩm.

Lập trình viên nên đọc vì những phiên bản Raspberry Pi 4 này có giá thấp hơn đáng kể nhưng vẫn hoạt động ổn định, giúp bạn có thể thử nghiệm các ứng dụng hoặc dự án với chi phí tiết kiệm mà không lo về hiệu suất bị giảm đáng kể.