Google logo

Google Engineering Blog & Tech Articles

Global technology leader developing innovative software, cloud services, and AI solutions across mobile, web, and enterprise platforms

570 engineering articles, tutorials, and technical insights from Google's engineering team

Latest Articles

Filter:
Google logo
Google
Advanced
The newly launched AI-Hypercomputer/tpu-codesign guide provides hardware-grounded design rules to help developers maximize performance on TPUs for LLMs and Mixture-of-Experts (MoE) workloads. Key principles include aligning reduction and output dimensions to 256 for systolic arrays, adopting LatentMoE to compress routed token traffic, and mapping parallel axes directly onto physical links.
Gagik Amirkhanyan, Vaibhav Singh
15 min read
Includes Code
--
Google logo
Google
Advanced
How we eliminated floating-point train–inference logprob drift for on-policy LLM reinforcement learning across 64 Cloud TPU v5p chips using a generalizable four-part JAX and Pallas recipe in Tunix.
Yuxuan Zhang, Lin Chai, Lance Wang, Tianshu Bao, Weiren Yu
7 min read
Includes Code
--
Google logo
Google
Advanced
This guide walks developers through setting up Agent Anomaly Detection on the Gemini Enterprise Agent Platform to monitor AI agent telemetry and catch behavioral risks. It demonstrates how to deploy a sample customer support agent, trigger out-of-scope security findings, and configure proactive alerts using Cloud Monitoring and Security Command Center.
Achuth Narayan Rajagopal, Derek Luan
10 min read
Includes Code
--
Google logo
Google
Intermediate
AI agent sessions remain active and idle for most of the workday, making traditional one-Pod-per-session allocations inefficient for cluster resources. By using open-source tools like Agent Substrate and AX on GKE, developers can park idle sessions in object storage to release compute resources while maintaining session state. This approach optimizes resource utilization and reduces Kubernetes node costs by up to 40%.
Drake Williams, Ivan Nardini
9 min read
Includes Code
--
Google logo
Google
Advanced
How pairing Go's surviving compiler bounds-check traps with Gemini 4 Argon and the Z3 constraint solver catches latent out-of-bounds panics in Go code and uncovers a slice-capacity blind spot in go test -fuzz.
Karl Weinmeister
12 min read
Includes Code
--
Google logo
Google
Intermediate
Agent Anomaly Detection (AAD) provides asynchronous, low-latency monitoring for AI agents by analyzing OpenTelemetry traces and logs to flag goal hijacks and tool misuse. Tested across five public security benchmarks, AAD successfully caught 87% of harmful instructions carried out by agents without adding runtime latency. Developers can integrate AAD alongside semantic governance and strict tool limits as part of a defense-in-depth strategy to secure agentic applications.
Nakul Javali
8 min read
Includes Code
--
Google logo
Google
Beginner
A2UI 1.0 is an open protocol for generative UI that enables AI agents to render interactive, native screens across web, mobile, and desktop applications. Developers can leverage open-source renderer SDKs—including React, Angular, Lit, Flutter, SwiftUI, and Jetpack Compose—alongside agent SDKs for Python, TypeScript, and Dart. The release introduces bidirectional function calls, swappable inference formats, and secure data binding to help teams seamlessly integrate generative UI into their product stacks.
Vishal Motwani, Google A2UI Team
8 min read
Includes Code
--
Google logo
Google
Advanced
Google AI Edge Gallery now supports on-device, offline Automatic Speech Recognition (ASR) and Text-to-Speech (TTS) capabilities for building developer-friendly voice applications. Developers can leverage ML Kit for optimized Android performance, or use MediaPipe Tasks and LiteRT for cross-platform open-weight model integration. This local pipeline seamlessly pairs lightweight speech models with on-device small language models like Gemini Nano and Gemma 4 to enable real-time streaming transcription and low-latency synthesis.
Google AI Edge Team, Android ML Team
9 min read
--
Google logo
Google
Intermediate
Pathways is a single-controller distributed training system designed to coordinate large-scale model training across multiple accelerator groups by representing computations as a global dataflow graph. Unlike traditional SPMD models like DDP and FSDP that rely on synchronous collectives, Pathways uses asynchronous dispatch and a central scheduler to maximize hardware utilization and prevent deadlocks in complex workloads. By decoupling model definition from physical hardware via virtual devices, developers can efficiently scale complex training architectures and seamlessly handle dynamic workloads like reinforcement learning post-training.
Sujeeth Jinesh, Shaurya Gupta, Luke Baumann, Monica Song, Ivan Nardini, Akanksha Gupta, Anthony Owusu, Jing Yang, Sadi Kneipp, Yu Li
13 min read
--
Google logo
Google
Beginner
Google Cloud's MaxText framework now supports long-context training up to 14 million tokens on Cloud TPUs by implementing four context-parallel strategies: All-gather, A2A, P2P, and USP. Powered by the Tokamax Splash Attention kernel, these strategies effectively manage attention compute and memory bottlenecks without materializing full key-value sequences in high-bandwidth memory. Developers can adopt these capabilities in MaxText v0.2.4 to achieve high Model FLOPs Utilization (MFU) across large-scale model training workflows.
Michael Tran, Nuojin Cheng, Deepak Patil
10 min read
--
Google logo
Google
Beginner
Developers can now build applications powered by Google Maps Platform directly within Google AI Studio by prompting Gemini with real-world geospatial concepts. The integration eliminates traditional configuration hurdles by providing one-click demo keys that generate instantly without requiring Cloud Console redirects or credit cards. This setup grants developers immediate access to core APIs—including Maps, Places, and Routes—supported by daily sandbox quotas for seamless prototyping.
Abinaya Selvam, Mike Taylor-Cai
3 min read
--
Google logo
Google
Advanced
Analysis of over 913,000 coding agent sessions reveals that 13% expose agents to inline TODO(b/...) comments. In 6% of encounters, agents act on these comments, with 83% being entirely unprompted by the developer. These detours result in significant overhead—adding 70% more steps, 77% more tokens, and 105% more wait time per session—split between chasing closed tickets (41%) and unauthorized scope creep on open bugs (57%). Higher-capability models show higher distraction rates, highlighting the need for automated cleanup and linters to maintain agent-friendly codebases
Sam Saccone, Madhuri Garikapati
5 min read
--
Google logo
Google
Intermediate
Per-token API prices keep falling, yet engineering bills for autonomous coding agents keep climbing....
Sergio Villani
11 min read
Includes Code
--
Google logo
Google
Intermediate
See Google A2Ui in action! Learn how to natively render agentic UIs across Langchain, AWS Strands, Microsoft agents, and Google ADK using CopilotKit SDKs for A2UI.
Google A2UI Team, Atai Barkai
8 min read
Includes Code
--
Google logo
Google
Advanced
This post walks through the Agents CLI evaluation loop for systematically improving agent performance through continuous testing and grading. It demonstrates how to catch prompt regressions, evaluate model backend swaps using side-by-side trace comparisons, and automate evaluation passes. Developers will also learn how to integrate nightly scheduled runs with threshold-based alerting to track quality over time.
Pier Paolo Ippolito, Elia Secchi
13 min read
Includes Code
--
Google logo
Google
Beginner
LiteRT in Google Play services now offers GPU inference powered by ML Drift, delivering about 1.4x faster performance alongside the new CompiledModel API. Switching from bundled binaries reduces app size by approximately 8 MB while enabling seamless, automatic runtime updates across billions of devices. Developers can easily adopt these updates by updating their dependencies and utilizing the new Android Studio LiteRT skill for automated migration.
Google AI Edge Team, Android ML Team
5 min read
Includes Code
--
Google logo
Google
Beginner
A2UI now supports swappable inference formats, introducing A2UI Express which uses up to 3x fewer tokens and cuts latency by compiling compact Python-style syntax into standard A2UI JSON via Agent SDKs. Developers can also use controlled generation with TypeSafe's Jev system-decision models to render known-shape screens in under 300 ms while smaller models stream text. These updates require no changes to frontend renderers or component catalogs, offering flexible performance levers for generative UI development across Python, TypeScript, and Dart.
Vishal Motwani, Google A2UI Team
9 min read
Includes Code
--
Google logo
Google
Advanced
This document provides a publication template and metadata structure for a technical blog post focused on building Conversational User Interfaces (CUIs) using AI. It outlines key developer focus areas, product stacks, and approval workflows required for publishing technical content through Google Developer Studio. The framework helps engineers organize their guides and solutions to effectively share AI-driven conversational interface designs with the developer community.
Prateek Khushalani
18 min read
Includes Code
--
Google logo
Google
Beginner
Google Workspace now offers native support for Markdown (.md) files, allowing developers and AI agents to seamlessly collaborate on documentation directly in Google Docs without manual file conversions. This integration preserves native file fidelity for use with Git repositories and developer tools while unlocking standard Docs features like line-level commenting, rendered Drive previews, and tracked version history.
Sophia Siao, Chelsea Rice
2 min read
--
Google logo
Google
Advanced
Learn how Project Chrysalis combines Google Cloud's Gemini Enterprise Agent Platform, Agent Development Kit (ADK), BigQuery, and a four-gate defense-in-depth security stack (Model Armor, Cloud KMS, Binary Authorization, and OPA) to build governed, self-evolving AI agents that remediate live production outages in 1.8 seconds.
Google logo
Google
Intermediate
API Gateway now supports real-time streaming—including Server-Sent Events, WebSockets, gRPC, and chunked transfer encoding—in Public Preview without requiring changes to your existing API specifications. This update allows developers to route long-lived, bi-directional streams and AI workloads directly through the gateway while maintaining consistent authentication, rate limiting, and observability. Additionally, gateways now feature deterministic default hostnames, simplifying environment provisioning and infrastructure-as-code management.
Google logo
Google
Advanced
The Agents CLI Extension System enables developers to use frameworks like LangChain, LangGraph, or custom enterprise stacks while retaining a unified deployment and evaluation workflow. It details a first-party LangChain template that scaffolds, runs, and deploys agents without requiring the Agent Development Kit (ADK). Additionally, it demonstrates how platform teams can author custom extensions to enforce policies such as automated software bill of materials (SBOM) scans during deployment.
Pier Paolo Ippolito, Elia Secchi
20 min read
Includes Code
--
Google logo
Google
Advanced
Google Research has completed the 32-bit open-source Coral NPU RISC-V architecture with the introduction of its Matrix Execution Unit, which delivers up to 1 TOPS while targeting 10 mW average power for edge inference. Alongside this architecture, the release of the v1.0 open-source IREE/MLIR compiler allows developers to take models from JAX, PyTorch, and TensorFlow straight to Coral NPU hardware without writing custom assembly. Developers can now prototype and deploy these multimodal edge applications using new hardware options like the Synaptics Astra SR200 MCU series and globally available Coralboard devkits.12
Billy Rutledge
7 min read
--
Google logo
Google
Advanced
This post announces native PyTorch backends for vLLM and SGLang powered by TorchTPU, enabling developers to run and optimize serving workloads using familiar PyTorch tools. Built in collaboration with open-source project maintainers, these engines include optimized TPU kernel implementations, production-ready features like KV cache management with TPU Sync, and disaggregated serving. Public benchmarks demonstrate state-of-the-art performance and industry-leading cost efficiency on Ironwood TPU across long-context and agentic workloads.
Qi Zhou, Chris Chan
7 min read
Includes Code
--
Google logo
Google
Advanced
Google AI Edge Model Explorer now supports custom compiler intermediate representations and open-source adapters like TOSA, VGF, and PTE for on-device machine learning. Developers can leverage hierarchical graph visualization, split-pane views, and per-node trace overlays to catch export regressions and verify hardware delegation. These tools enable engineering teams to transparently debug execution graphs and pinpoint cycle-level latency bottlenecks locally.
Eric Yang, Alan Ma
4 min read
Includes Code
--
Google logo
Google
Advanced
TorchTPU integrates a native torch.device("tpu") backend directly into PyTorch, allowing developers to train large-scale models on Google Cloud TPUs without rewriting code in JAX. The TorchTitan TPU fork enhances this with production-grade scaling primitives like SimpleFSDP, Graph Trainer, and custom Pallas kernels that achieve high Model Flops Utilization (MFU) across dense and MoE architectures. Developers can leverage these battle-tested optimization playbooks, recipes, and the Hugging Face modeling backend to efficiently scale their own workloads on TPU infrastructure.
Aleksey Vlasenko, Rob Torres, Will Constable
22 min read
Includes Code
--
Google logo
Google
Intermediate
The open-source Agent Development Kit (ADK) and gemini-3.8-live enable developers to build responsive voice agents by implementing non-blocking tools in Python. Developers can use streaming tools to yield progress updates or background execution patterns to keep conversations fluid during long-running backend queries. Additionally, configuring response scheduling behaviors such as WHEN_IDLE, INTERRUPT, or SILENT gives developers precise control over when and how tool results are delivered into the conversation.
Stephen Allen, Kaz Sato
9 min read
Includes Code
--
Google logo
Google
Intermediate
Meet the LiteRT Tensor API: giving developers native C++ control over the model graph to unify end-to-end ML computations on heterogeneous edge hardware.
Ping Yu, Quentin Khan, Chintan Parikh, Lu Wang
10 min read
Includes Code
--
Google logo
Google
Intermediate
A practical engineering guide and reference architecture for establishing continuous evaluation pipelines for multi-turn AI agents using Gemini Enterprise Agent Platform, Google ADK, and OpenInference standards.
Hugo Selbie, Dani Zamora, Jesse Carah
12 min read
Includes Code
--
Google logo
Google
Intermediate
The google/adk-recipes repository unifies Agent Development Kit (ADK) resources into three structured layers—core/, plugins/, and contrib/—serving as a single source of truth for building AI agents. Developers can leverage reference patterns, install modular vertical skills (such as retail product search and virtual try-on) directly into their coding assistants, and deploy production-ready services across multiple languages. This cohesive structure streamlines the entire workflow from initial scaffolding to full deployment on Google Cloud infrastructure.
Pier Paolo Ippolito, Shahin Saadati
14 min read
Includes Code
--
Google logo
Google
Intermediate
Google OAuth Platform now supports DPoP (Demonstrating Proof-of-Possession) to sender-constrain refresh tokens, neutralizing Infostealer malware and adversary-in-the-middle attacks. By cryptographically binding long-lived refresh tokens while maintaining short-lived access tokens as standard Bearer tokens, developers achieve robust credential security without incurring high-QPS API performance overhead. Additionally, developers integrating multi-agent or Model Context Protocol (MCP) workloads can use Google Cloud Secure Token Service (STS) to handle secure token delegation across agent chains.
Sergei Akulich, Brian Daugherty
5 min read
--
Google logo
Google
Intermediate
Apigee hybrid 1.17.0 introduces native Model Context Protocol (MCP) support, enabling the MCP data plane to run directly within your own Kubernetes cluster. This architecture ensures that AI agent request traffic and tool execution remain entirely inside your infrastructure while pulling configuration updates from the control plane. Developers can expose OpenAPI 3.0.x specifications as discoverable MCP tools, leverage existing security policies and API products, and trace end-to-end requests using W3C trace context headers.
Ayo Salawu, Sundar Ramamoorthy
8 min read
Includes Code
--
Google logo
Google
Intermediate
Gemini 4 Argon is a general-purpose physical reasoning model designed to enhance workflows across the robotics stack with advanced capabilities in physical understanding, code generation, and visual-semantic reasoning. Developers can access and integrate Argon across Google's developer platforms—including Google AI Studio, the Gemini API, and Google Antigravity—to build and test sophisticated robotic applications. By leveraging simulation tools like MuJoCo and code-as-policies, engineers can seamlessly transition from real-world data or sketches to physically verified robot control and self-improvement workflows.
Kanishka Rao
18 min read
Includes Code
--
Google logo
Google
Intermediate
How our engineering team ran Gemini 4 Argon unattended overnight with .agents/skills and /goal mode to cut in-memory session prep overhead by 40% in long sessions and speed up cold imports by 21% in Google's open-source Python Agent Development Kit (ADK).
George Weale, Bo Yang, Andrew Biernat
10 min read
Includes Code
--
Google logo
Google
Advanced
Google AI Edge has introduced ML Drift, a high-performance, universal GPU compute framework designed to accelerate on-device AI and machine learning inference. By abstracting the complexities of hardware and low-level APIs—including OpenGL ES, OpenCL, Metal, and WebGPU—ML Drift enables developers to build real-time, interactive ML experiences, from sophisticated video effects to generative AI, across mobile and desktop platforms.
Chintan Parikh, Juhyun Lee
9 min read
--
Google logo
Google
Intermediate
How do you keep improving an AI agent after the first 80%, once it serves real production traffic and silent quality regressions return HTTP 200 OK? We are open-sourcing AQuA (Ambient Quality Agent), a reference implementation that runs beside your production agent in Google Cloud, sweeps and verifies failure clusters against conversation transcripts, and diagnoses root causes anchored to the exact deployed source snapshot.
Dima Melnyk, Elia Secchi
18 min read
Includes Code
--
Google logo
Google
Beginner
Google is launching the Developer Knowledge API and MCP Server in public preview. This new toolset provides a canonical, machine-readable way for AI assistants and agentic platforms to search and retrieve up-to-date documentation across Firebase, Google Cloud, Android, and more. By using the official MCP server, developers can connect tools directly to Google’s documentation corpus, ensuring that AI-generated code and guidance are based on authoritative, real-time context.
Christina Gonzalez
5 min read
Includes Code
--
Google logo
Google
Intermediate
EmbeddingGemma 2 is a compact, open-source multimodal embedding model that maps text, code, images, video, and audio into a unified 768-dimensional space. Developers can use the sentence-transformers library to selectively load modular modality encoders—ranging from 270M to 740M parameters—to optimize memory usage. Additionally, Matryoshka Representation Learning enables dynamic dimension truncation down to 128d, significantly reducing vector database storage requirements while maintaining high retrieval performance.
Maarten Grootendorst, Ian Ballantyne
7 min read
Includes Code
--
Google logo
Google
Advanced
EmbeddingGemma 2 is a new 740M open-weight multimodal model that maps text, images, video, and audio into a unified vector space for privacy-first, on-device retrieval. Developers can easily integrate these capabilities cross-platform using MediaPipe Tasks or optimize fine-grained performance across CPU, GPU, and NPU accelerators with LiteRT. The model enables ultra-low-latency local solutions like search-as-you-type media retrieval, keyframe video moments finding, and zero-shot intent routing.
Google AI Edge Team, Android ML Team
11 min read
Includes Code
--
Google logo
Google
Advanced
To address the quadratic latency bottleneck of self-attention in high-resolution video diffusion models, developers implemented Sparse VideoGen (SVG) to dynamically route attention heads to highly structured spatial or temporal sparse masks. Translating this algorithmic sparsity into physical hardware speedups on TPUs required optimizing the Splash Attention kernel by bypassing empty memory tiles, restricting exact coordinate masking strictly to boundary tiles, and permuting token memory layouts into temporal-major order for contiguous access. By aligning these sparse masks with actual hardware tile execution, the combined optimizations significantly reduced wasted matrix operations and achieved up to a 1.69x end-to-end inference speedup for 1440p video generation.
Ravisri Valluri, Sagar Chapara, Rishabh Manoj
13 min read
--
Google logo
Google
Advanced
The MaxText team successfully reproduced AI2’s OLMo 3 7B language model from scratch on Google Cloud TPUs using JAX/XLA, precisely matching the original PyTorch-on-GPU reference across pre-training and mid-training stages on all held-out evaluations. The implementation achieved up to 57.4% Model Flops Utilization (MFU) and demonstrated robust infrastructure portability by surviving mid-run cluster resizes and cross-generation TPU shifts without requiring recipe alterations. Crucially, the exercise proved the necessity of comprehensive held-out validation by catching a silent data-loader memorization bug that artificially depressed training loss and would have otherwise faked a performance win.
Gagik Amirkhanyan, Ran Ran, Aireen Mei, Matt Davidow, TPU Inference Software Engineering Team, Google Cloud, AI2 Team
26 min read
Includes Code
--
Google logo
Google
Intermediate
Google Cloud API Gateway now acts as a native remote Model Context Protocol (MCP) server, eliminating the need to build and maintain custom middleware to expose REST APIs to AI agents. By simply adding specific annotations (like x-google-api-management.mcp) to existing OpenAPI 3.x specifications, developers can instantly convert standard REST operations into discoverable, agent-ready tools. The gateway automatically transcodes incoming MCP JSON-RPC requests into REST calls, ensuring that your existing authentication, quotas, and logging policies apply seamlessly to agent traffic without requiring new infrastructure.
Sanjay Pujare, Paul Howell, Geir Sjurseth
5 min read
Includes Code
--
Google logo
Google
Beginner
The Google Antigravity SDK now empowers developers to execute offline, agentic workflows locally using models like Gemma 4 26B A4B via LiteRT. This update facilitates powerful hybrid orchestration architectures, allowing a cloud model to act as a lightweight planner while local models securely handle token-intensive tasks—like code auditing and patching—directly on-device. Furthermore, the SDK provides drop-in support for OpenAI-compatible inference servers like Ollama and vLLM, enabling the seamless creation of privacy-first, autonomous local utilities.
Sachin Kotwani, Taylor Mullen
5 min read
Includes Code
--
Google logo
Google
Beginner
Unlock premium Google Colab compute with Google AI. Subscribers now get priority accelerators, Premium GPUs, and background execution for long training runs.
Spencer Shumway, Niv Govindaraju, Mae LaPresta
1 min read
--
Google logo
Google
Intermediate
Google has partnered with Speakeasy to open-source their OpenAPI code generation suite under the AGPLv3 license, a strategic move prompted by the sudden shutdown of Google's previous proprietary SDK provider. The newly open-sourced suite equips developers with deterministic, multi-language SDK generators that natively support strict typing and SSE streaming, alongside tools for compiling agent-native CLIs and documentation MCP servers. Engineering teams can now safely integrate this robust tooling directly into their CI pipelines to automatically generate reliable client libraries for their own APIs, all while retaining complete licensing control over the output code.
Amir Hardon, Philipp Schmid
3 min read
Includes Code
--
Google logo
Google
Intermediate
Agent Anomaly Detection is a new, out-of-band oversight layer for the Gemini Enterprise Agent Platform that analyzes OpenTelemetry traces and tool calls to catch behavioral risks without adding runtime latency to live requests. It utilizes a multi-tiered detection pipeline—combining lightweight statistical scanning with deep LLM-based reasoning—to identify logical anomalies and policy violations grounded in the OWASP Agentic Top 10. Developers can triage these automated findings within Security Command Center or leverage the exposed API to programmatically block subsequent tool calls when an agent breaches defined risk thresholds.
Achuth Narayan Rajagopal
5 min read
Includes Code
--
Google logo
Google
Advanced
This blog post explores how to transition AI agents from static, build-time security controls to dynamic runtime governance using the Gemini Enterprise Agent Platform. It highlights three primary managed defenses: Model Armor for screening edge prompts, Semantic Governance Policies for evaluating tool intent against business rules, and Agent Anomaly Detection for catching multi-turn exploits. By shifting these capabilities to the platform level, security administrators can dynamically enforce policies and neutralize complex attacks without needing to modify or redeploy the agent's underlying code.
Eric Dong, Shubham Saboo
12 min read
Includes Code
--
Google logo
Google
Advanced
Imagine going to sleep after writing a single Markdown specification and waking up to find that an A...
Wei Wei
4 min read
Includes Code
--
Google logo
Google
Intermediate
While end-to-end benchmarks like SWE-bench provide broad performance scores for AI agents, they are often expensive, slow, and lack the root-cause diagnostics needed to explain exactly where an agent's logic broke down. To solve this, developers should adopt behavioral evaluations—fast, local, unit-style tests that assert on discrete intermediate actions, such as verifying specific tool calls or file modifications rather than final string equality. By building these inexpensive micro-checks alongside macro benchmarks, engineering teams can confidently iterate on system prompts and upgrade models without the risk of regressions.
Taylor Mullen, Christian Gunderman
6 min read
Includes Code
--
Google logo
Google
Intermediate
Google has officially released version 1.0 of the Agent Development Kit (ADK) for Kotlin, achieving full feature parity with the Python and Java ADK cores to enable idiomatic, multi-agent AI development. Built on Kotlin Multiplatform (KMP), the framework leverages Kotlin Symbol Processing (KSP) for zero-reflection, type-safe function calling, alongside advanced orchestration capabilities like human-in-the-loop workflows and context compaction. Additionally, the release introduces a robust suite of Android-first extensions, allowing mobile developers to integrate local models via LiteRT-LM, cloud reasoning through Firebase AI, session persistence using Room, and semantic memory powered by AppSearch.
Guillaume Laforge
9 min read
Includes Code
--