# News: August 27, 2026 > 16 curated links on AI agents, Python, security, Google Cloud, and engineering, selected by Médéric Hurier. - Author: [Médéric Hurier (Fmind)](https://www.fmind.dev/) - Published: 2026-08-27 - URL: https://www.fmind.dev/news/2026-08-27/ - [GLM-5.3-Flash](): Zhipu AI releases GLM-5.3-Flash, an open-weight 320B parameter multimodal model (18B active) featuring hybrid sparse-linear attention, a 1-million-token context window, and native GUI agent capabilities. - [Qwen3.8-Flash-Next](): Alibaba introduces Qwen3.8-Flash-Next, an open-weight architectural preview for Qwen4 featuring hybrid Gated DeltaNet attention, dynamic gated residuals, and N-gram embeddings across 125B parameters. - [Intelligent Transcription with Gemini 3.5 Transcribe](): Google unveils Gemini 3.5 Transcribe and Gemini 3.5 Transcribe Live, delivering intent-aware speech-to-text transcription with real-time streaming, filler-word filtering, and multilingual support across 85 languages. - [Flexible Billing and Cost Controls for Agents on Google Cloud](): Google Cloud introduces flexible billing options and granular cost management controls for autonomous agent workloads across Agent Platform, Gemini Enterprise, Google Antigravity, and Android Studio. - [Best Practices for Dynamic Capacity Management](): Google Cloud shares architectural best practices for dynamic capacity management to optimize compute utilization, minimize idle accelerator resources, and scale AI workloads efficiently. - [Scale OKF Bundles Across an Organization with Knowledge Catalog](): Google Cloud details how Knowledge Catalog enables organizations to govern, share, and scale Open Knowledge Framework (OKF) data bundles securely across teams. - [Build a Real-Time Voice AI Agent with Gemini Live API](): Google Cloud Tech demonstrates how developers can build bidirectional, real-time voice AI agents with native interruption handling and tool invocation using the Gemini Live API. - [TechByte: AI Infrastructure in 2026 - Architecting for Agentic AI](): Google Cloud Events hosts a technical session exploring how enterprise infrastructure must evolve to support autonomous agentic AI architectures while optimizing inference cost and performance. - [Your Chance to Start Building AI Agents from the Absolute Basics](): Google Cloud announces Agent Valley, a free five-week live learning series designed to teach developers foundational principles and techniques for building AI agents. - [Nvidia Forecasts 70% Revenue Growth on Unabated AI Compute Demand](): Nvidia forecasts a 70% surge in revenue for the next fiscal year, reflecting persistent global enterprise demand for AI acceleration hardware. - [Dylan Patel – Two Labs Will Soon Control Most of the World's Workforce](): SemiAnalysis founder Dylan Patel joins Dwarkesh Patel to analyze frontier lab economics, shifting compute allocations from training to inference, and the concentration of global AI FLOPs. - [Turn Your Voice into Action with New Productivity Features in Gemini Live](): Google rolls out productivity upgrades to Gemini Live, enabling users to delegate everyday tasks and manage personal workflows hands-free using real-time voice interactions. - [AI Coding Will Prevent Expertise](): Lars Faye argues that over-relying on AI coding assistants removes the essential cognitive friction needed for early-career software engineers to develop deep technical expertise. - [Why the FBI Gave Up](): The Infographics Show examines the shift in U.S. cyberwarfare strategy as authorities consider authorizing private corporations to conduct offensive operations against foreign hackers. - [Google’s New Robot Will Steal Your Job](): SAMTIME delivers a satirical critique analyzing Google's latest Gemini Robotics demonstrations and the broader hype surrounding humanoid automation in the workplace. - [How Americans See EU Tech](): Kai Lentit presents a humorous parody satirizing the cultural and regulatory differences between European tech founders and American Silicon Valley startup ecosystems.