'%3e%3crect%20x='16'%20y='51'%20width='68'%20height='27'%20rx='13.5'/%3e%3ccircle%20cx='32'%20cy='53'%20r='15'/%3e%3ccircle%20cx='53'%20cy='43'%20r='20'/%3e%3ccircle%20cx='71'%20cy='57'%20r='12'/%3e%3c/g%3e%3c/svg%3e){kind=small}
Cloud ChangelogDaily digest · Thursday, November 12, 2026
Gemini 4 Hydrogen GA, FSx for Lustre Intelligent-Tiering 75% cheaper entry, AgentCore next-gen runtime default
69 announcements · 29 GA · 12 previews · 1 price change · 1 Region expansion
Top highlights
Google Gemini API · Gemini API · AI and machine learning
Adds xhigh thinking level; $1.50 per 1M input and $7.50 per 1M output tokens, free tier available
AWS · Amazon Bedrock · AI and machine learning
Sessions up to 8 vCPU and 32 GB; start under 2 seconds at P75 in 19 Regions
AWS · Amazon FSx for Lustre · Storage
Minimum throughput falls from 4,000 to 1,000 MBps, metadata from 6,000 to 1,500 IOPS
Video summary 1:32
WatchWatch on YouTube(opens a new tab)AWS17
AWS · Amazon EC2 · Compute and containers
Console chat panel uses DevOps Agent to diagnose up to 50 failed pipelines and create resources
AWS · Amazon Cognito · Security, identity and compliance
Supports OAuth 2.0 token exchange (RFC 8693) for narrower, audience-bound downstream tokens
AWS · Amazon Connect · Application integration
Agents can merge related cases into one and link related cases in agent workspace
New region: SNS push notifications in 4 more Regions
Analytics 4
AWS · Amazon MWAA · Analytics
Fine-grained access control via IAM Identity Center and multi-team environments
AWS · AWS Glue · Analytics
Apache Spark jobs upgrade to latest Glue runtime when version is set to Auto, with validation
AWS · Amazon Managed Service for Apache Flink · Analytics
Enhanced snapshots capture, export and restore application state across applications and Regions
AWS · Amazon MWAA · Amazon SageMaker · Analytics
Supports DuckDB operators for in-memory analytics on S3 data in workflows, including SageMaker Unified Studio
Storage 3
AWS · Amazon FSx for Lustre · Storage
Minimum throughput cut from 4,000 to 1,000 MBps, metadata from 6,000 to 1,500 IOPS
AWS · AWS Backup · Amazon Elastic Block Store · Storage
Supports archiving Amazon EBS snapshots to EBS Snapshots Archive
AWS · AWS Backup · Amazon EKS · Storage
Resource modifiers transform Kubernetes resources during Amazon EKS restores using a YAML file
AI and machine learning 2
AWS · Amazon Bedrock · AI and machine learning
Next generation runtime is default for new runtimes in 19 Regions, with sessions up to 8 vCPU and 32 GB
AWS · Amazon QuickSight · AI and machine learning
Admins restrict creating, updating or sharing data sources by type through custom permissions
Databases 2
AWS · Amazon Aurora · Databases
Up to 30% faster than Graviton4 on Aurora and RDS, up to 192 vCPUs
AWS · Amazon RDS for SQL Server · Databases
M8id and R8id instances with local NVMe storage for tempdb, up to 384 vCPUs
Networking and content delivery 2
AWS · AWS Direct Connect · AWS Cloud WAN · Networking and content delivery
BGP communities supported on Direct Connect attachments for routing policy
AWS · Amazon Route 53 · Networking and content delivery
Self-service restore of expired and deleted domains through console, CLI or SDKs
Google Cloud30
Google Cloud · Cloud Service Mesh · Networking and content delivery
Versions 1.31.3-asm.9, 1.30.7-asm.9 and 1.29.10-asm.4 fix vulnerabilities in GCP-2026-080
Serverless 14
Google Cloud · App Engine standard environment Go · Serverless
Direct VPC egress is GA; active versions can deploy to Cloud Run from Migration hub
Google Cloud · App Engine standard environment Java · Serverless
Direct VPC egress is GA; active versions can deploy to Cloud Run from Migration hub
Google Cloud · App Engine standard environment Node.js · Serverless
Direct VPC egress is GA; active versions can deploy to Cloud Run from Migration hub
Google Cloud · App Engine standard environment PHP · Serverless
Direct VPC egress is GA; active versions can deploy to Cloud Run from Migration hub
Google Cloud · App Engine standard environment Python · Serverless
Direct VPC egress is GA; active versions can deploy to Cloud Run from Migration hub
Google Cloud · App Engine standard environment Ruby · Serverless
Direct VPC egress is GA; active versions can deploy to Cloud Run from Migration hub
Google Cloud · App Engine flexible environment custom runtimes · Serverless
Migration hub checks compatibility and migrates custom runtime services to Cloud Run
Google Cloud · App Engine flexible environment Go · Serverless
Migration hub checks compatibility and migrates Go services to Cloud Run
Google Cloud · App Engine flexible environment Java · Serverless
Migration hub checks compatibility and migrates Java services to Cloud Run
Google Cloud · App Engine flexible environment .NET · Serverless
Migration hub checks compatibility and migrates .NET services to Cloud Run
Google Cloud · App Engine flexible environment Node.js · Serverless
Migration hub checks compatibility and migrates Node.js services to Cloud Run
Google Cloud · App Engine flexible environment PHP · Serverless
Migration hub checks compatibility and migrates PHP services to Cloud Run
Google Cloud · App Engine flexible environment Python · Serverless
Migration hub checks compatibility and migrates Python services to Cloud Run
Google Cloud · App Engine flexible environment Ruby · Serverless
Migration hub checks compatibility and migrates flexible services to Cloud Run
Compute and containers 4
Google Cloud · Batch · Compute and containers
Jobs can use TPU7x, v6e and v5p TPU VMs for ML and AI workloads
Google Cloud · Google Kubernetes Engine · Compute and containers
Runs isolated, stateful, single-replica workloads such as AI agents at no extra cost; Agent Gateway enters Preview
Google Cloud · Compute Engine · Compute and containers
Override template disk and minimum CPU platform per instance selection
Google Cloud · Container Optimized OS
Adds RTC support in A4X-Max images and Lustre client driver 2.16 package
Analytics 3
Google Cloud · BigQuery · Analytics
Table-level data lineage tracking in Knowledge Catalog
Google Cloud · Dataform
Table-level data lineage tracking in Knowledge Catalog for workflows and BigQuery pipelines
Google Cloud · Looker · Analytics
Deploying in Rapid and No Channel; Regular channel stays at 26.20
AI and machine learning 2
Google Cloud · Gemini Enterprise Agent Platform · AI and machine learning
Unified 1,048,576-token limit, four thinking levels, implicit caching, Office and OpenDocument inputs
Google Cloud · Gemini Enterprise · AI and machine learning
Gemini 3.8 Flash available in Canada and Japan regions; skill owners can restore earlier skill versions
Security, identity and compliance 2
Google Cloud · API Keys API
Lets AI applications create, inspect, restrict and manage lifecycle of API keys in projects
Google Cloud · Model Armor · Security, identity and compliance
Filter v4 raises recall for prompt injection, tool calls and jailbreak attacks
Management, governance and cost 2
Google Cloud · Cloud Logging
Alerting policies on Observability Analytics query results, created in console or Cloud Monitoring API
Google Cloud · Cloud Trace
Alerting policies on Observability Analytics query results, created in console or Cloud Monitoring API
Application integration 2
Google Cloud · Apigee hybrid · Application integration
SanitizeUserPrompt and SanitizeModelResponse policies handle LLM function calling flows with new XML elements
Google Cloud · Apigee hybrid · Application integration
Announcement with no details provided
Azure4
Azure · Azure SQL Managed Instance · Databases
Microsoft-managed performance telemetry with resource-level and at-scale dashboards, plus KQL queries via…
Azure · Azure Kubernetes Service (AKS) · Azure Monitor · Compute and containers
Ingestion limits scale automatically per workspace based on observed usage patterns
Security, identity and compliance 2
Azure · Azure Front Door · Networking and content delivery
Client certificate authentication on Premium with four validation modes and root-CA-only trust configuration
Azure · Azure Front Door · Networking and content delivery
JWT validation of Microsoft Entra ID bearer tokens at the edge on Premium, with Block and Log modes
Cloudflare11
Cloudflare · D1 · Databases
Databases with virtual tables, such as FTS5 and RTree, can be exported via SQLite CLI-only dump
Cloudflare · Cloudflare CLI · Workers · Developer tools, web and mobile
CLOUDFLARE_API_HEADERS variable adds custom headers to Cloudflare API requests
AI and machine learning 4
Cloudflare · Workers AI · AI and machine learning
Multimodal MoE model with 594B parameters and 1,048,576 token context, on Workers AI
Cloudflare · AI Search · D1 · AI and machine learning
Indexes rows from a D1 database via a read-only query, with scheduled syncs
Cloudflare · Workers AI · AI and machine learning
Open multimodal embedding model maps text, images, video and audio into 768-dimension vectors
Cloudflare · Browser Run · AI and machine learning
Browser engine for AI agents released on GitHub under Apache 2.0; runs on Workers
Security, identity and compliance 3
Cloudflare · Cloudflare Tunnel · Cloudflare Tunnel for SASE · Security, identity and compliance
Normalization handles semicolon-delimited path parameters before ingress matching; default on from 2027-02-08
Cloudflare · Rules · Security, identity and compliance
Adds sha1(), md5() and sha512(), returning raw bytes from strings or byte arrays
Cloudflare · WAF · Security, identity and compliance
Terraform provider 5.29.0 adds resources for enabling it and defining custom topics
Media, games and IoT 2
Cloudflare · Realtime · Media, games and IoT
Custom validity windows via not_before and expires_on, plus revoking a single token
Cloudflare · Realtime · Media, games and IoT
New npm package filters background noise in the browser for React, Angular and JavaScript web apps
Anthropic1
Anthropic · Claude Platform · AI and machine learning
Run the full Managed Agents service on your own infrastructure, storing agents and sessions there
Google Gemini API1
Google Gemini API · Gemini API · AI and machine learning
Adds xhigh thinking level; $1.50 per 1M input, $7.50 per 1M output tokens, free tier
Together AI2
Together AI · Together AI platform · AI and machine learning
Serverless model with 1,048,576-token context, NVFP4 quantization, text and image input, function calling, structured…
Together AI · Together AI platform · AI and machine learning
Added to serverless with 262,144 context, FP8, function calling, structured outputs
Cerebras1
Cerebras · Cerebras Inference · AI and machine learning
Added to Dedicated model catalog with image inputs and reasoning levels low, medium, high, max
TwelveLabs1
TwelveLabs · TwelveLabs API · AI and machine learning
Free plan now 10 requests/minute, 200/day; Plus and Pro 30/minute, 1,000/day
MiniMax1
MiniMax · MiniMax API · AI and machine learning
New open-weight image model released by MiniMax
From 69 announcements in the release notes and changelogs of AWS, Google Cloud, Azure, Cloudflare, Anthropic, Google Gemini API, Together AI, Cerebras, TwelveLabs and MiniMax.