What's new in documentation

Explore the new standalone documentation site for BigLake. Learn how BigLake unites Google Cloud and open source to build an open, managed, and high-performance lakehouse with built-in governance.

Version 1.12 of the ABAP SDK for Google Cloud adds support for Gemma models and gives developers more control over AI behavior.

Learn how to upload data files (CSV, XLS, XLSX) and instantly create Explores without LookML using the new Self-service Explores feature, now in Public Preview.

The Conversational Analytics documentation has been split into separate, dedicated documentation sets for Looker and Looker Studio.

For internal passthrough Network Load Balancers, you can now configure a lower minimum idle timeout of up to 60 seconds for all combinations of session affinity and connection tracking mode settings.

For additional layers of security and control, you can now use query templates to predefine and limit the queries that can be run in data clean rooms.

You can now add tables and views as tasks directly to your BigQuery pipelines to build more efficient, data-driven workflows. This feature is in preview.

Config Sync documentation has been refactored to help new users, including a new guide to install Config Sync with default settings, and updated pages for custom installations and authentication.

Deploy a Python-based agent to a production-ready GKE Autopilot cluster with GPU acceleration, using Google ADK and a self-hosted Llama 3.1 model served by vLLM.

Flex-start VMs run for up to seven days and help you acquire high-demand resources like GPUs at a discounted price for short-duration workloads.

Help secure your resources with Privileged Access Manager (PAM) advanced-tier features in Public Preview. Implement multi-level approvals, customize grant scopes, utilize service account approvals, and more for fine-grained temporary access control.

This comprehensive guide helps you design, deploy, and manage certificates using Google Cloud.

Control access to websites and webpages by blocking or allowing URLs, using domain and Server Name Indication (SNI) information in egress HTTP(S) messages.

Create and configure producer and consumer resources to set up out-of-band integration for Network Security Integration to inspect and monitor data.

Four new decision trees provide clear guidance for common Looker support questions, empowering customers to self-serve solutions.

Choose the best method for connecting to AlloyDB based on your workload environment, secure connectivity needs, and network topology.

Explore new tutorials for AI/ML workloads on AI Hypercomputer, including: vLLM on GKE to run inference with Llama 4 or Qwen3, fine-tune Gemma 3 on an A4 GKE cluster, use Ray to fine-tune Gemma 3 for vision tasks on GKE, and train Qwen2 on an A4 Slurm cluster.

As part of the Fast Onboarding program, standalone organizations are now automatically created for new Google Cloud customers upon sign-up, simplifying the onboarding process.

Deploy Gemini models on your own hardware connected to your local network with Gemini on Google Distributed Cloud connected API. This feature is in Public Preview.

Access updated product documentation for the 1.11.0 release of Google Distributed Cloud connected.

You can now visualize your geospatial query results on an interactive map in BigQuery Studio.

Define a central list of IP addresses that can be used in high-level rules to provide consistent control for your entire organization and reduce overhead for network and project owners.

By moving all our technical content to one, dedicated platform, we've created a unified foundation that makes it easier to build the next generation of AI-driven experiences.

NCCL/gIB is Google's enhanced version of the NVIDIA Collective Communications Library (NCCL) for GPU-to-GPU communication primitives.

This guide covers client-side and server-side factors and includes a flowchart to help you visualize the troubleshooting process.

Get deeper insights into cache performance and a clear overview of core Cloud CDN metrics, without custom configurations. Dashboards are enabled by default and give you a quick insights into system health

You can now enable the automatic selection of a processing location in your BigQuery pipeline configurations.

Cloud Run now supports popular Python web frameworks like FastAPI, Gradio, and Streamlit, making it easier than ever to deploy modern workloads.

Access updated product documentation for the 1.10.0 release of Google Distributed Cloud connected.

Deploy production-grade AI with the GKE Inference Quickstart tool. This guide focuses on performance and cost analysis, providing details on cost calculation, gcloud CLI commands, and performance optimization.

AlloyDB Omni product documentation now offers separate documentation for each deployment environment, making it easier to follow workflows.

GKE now offers a unified AI/ML workloads documentation suite, organizing content around user workflows like inference and training, and fully integrating documentation for Ray on GKE.

A4X VMs on Google Cloud use the NVIDIA GB200 Superchip and multi-node NVLink to offer exascale computing with up to 72 GPUs for large-scale AI and HPC workloads.

Configure listings for multiple regions for shared datasets and linked dataset replicas in BigQuery sharing. This feature is in preview.

Build enterprise-ready AI applications faster and with less complexity by bringing vector search, AI query engine, and natural language to your operational data.

Serve popular open LLMs on GKE for inference, using a pre-configured, production-ready reference architecture, with this new Terraform tutorial.

Accelerate app development and deploy intuitive chat experiences that let customers ask natural language questions about their relational data.

New to the app-centric world? Learn how App Hub, Application Design Center, and Cloud Hub help you manage, design, and operate your applications on Google Cloud.

Decide on the right Looker version for you. This comprehensive matrix compares features across Looker (Google Cloud core), Looker (original), and customer-hosted deployments.

The most recent changes to Google Cloud products.

The latest best practices and reference architectures for Google Cloud.

Join and learn from those who develop, deploy, and operate on Google Cloud.