Releasesby GlyphClickHouse release 26.10
ClickHouse 26.10 adds function composition, improved join flexibility, faster GROUP BY queries, and disk spilling for IN subqueries. It also includes an experimental columns cache and support for Freeform and SQLite formats.
Friday, November 6, 20263 releases
Enterprise Environments Page Loads for Large Enterprises
The environments page has been optimized to prevent timeouts for large enterprise organizations.
mujoco 4.0.0
MuJoCo 4.0.0 introduces speculative contacts to prevent tunneling of fast-moving objects and adds stateful elasto-plastic deformation for joints, tendons, and flexes. The Python distribution has also been reorganized into separate core, studio, and default packages.
Thursday, November 5, 20263 releases
Lower costs for streaming ingestion with Foundry worker
Compute costs for streaming ingests running on Foundry worker are reduced by around 97.5% starting November 1, 2026. This cost reduction builds on previous pricing updates to facilitate migration from the legacy agent worker to Foundry worker.
API server autoscaling is generally available
API server autoscaling for Airflow 3 deployments is now generally available. It automatically adjusts replicas based on CPU utilization.
Metrics API capacity errors | Enterprise
The CodeRabbit Metrics API now distinguishes between request rate limits and query capacity limits. Users should implement retries based on the Retry-After header when receiving 429 or 503 responses.
Wednesday, November 4, 20269 releases
Google Cloud4 products
Syncing data from BigQuery into AlloyDB is generally available ( GA )
Syncing data from BigQuery into AlloyDB is generally available ( GA ). This feature lets you copy or mirror analytical tables from BigQuery into AlloyDB for low-latency operational access using the alloydb_sync extension. For more information, see Sync BigQuery data to AlloyDB .
Compare AI agent changes with Online Experiments in Agent Observability
Datadog Online Experiments allows engineers to test AI agent changes against production traffic. It enables comparison of response quality, cost, latency, and error rates.
Zed v1.27.0
This week's release includes syntax highlighting for Go 1.27 generic methods, an option to open README files by default, and AI commit messages that follow your repository's recent commit style.
Ray-2.60.0
Ray 2.60.0 released shuffle v2 as default for joins and aggregations, updated Parquet task sizing from row-group metadata, added rolling update surge controls in Ray Serve, and patched security vulnerabilities.
Tuesday, November 3, 20262 releases
Inference New Serverless mode: Ultra-fast (Preview) Ultra-fast is a new high-speed Serverless mode, now in...
Inference New Serverless mode: Ultra-fast (Preview) Ultra-fast is a new high-speed Serverless mode, now in Preview, for highly latency-sensitive applications. Ultra-fast variants aim for 6-8x generation speed, and up to 10x for agentic coding tasks. Like Fast, it is not a different model, so output quality is the same.
Agents SDK now supports Pi 1.5 with fewer storage writes and lower memory use
Agents SDK v0.33.0 upgrades PiHarness to Pi 1.5 ↗︎ . The release reduces storage transactions while streaming, improves throughput across many sessions, and reduces memory use when streaming large tool calls. Performance improvements Far fewer storage writes while streaming.
Monday, November 2, 20265 releases
You can now tune SQL queries using AI
You can now tune SQL queries using AI. Available starting with version 1.6.2 of the google_ml_integration extension, the ai.tune_sql function uses an LLM along with query execution plans, schemas, and runtime stats to recommend query optimizations.
You can now monitor the performance, adoption, latency, and costs of your data agents and their conversations...
You can now monitor the performance, adoption, latency, and costs of your data agents and their conversations by using Google Cloud Observability in BigQuery. This feature is generally available (GA).
Amazon Aurora MySQL 3.14 (compatible with MySQL 8.0.46) is generally available
Starting today, Amazon Aurora MySQL-Compatible Edition 3 (compatible with MySQL 8.0) supports MySQL 8.0.46, which includes community MySQL fixes and Aurora-specific improvements.
Diagnose frontend performance issues with RUM Profiling
Datadog RUM Profiling is available to identify CPU-intensive code causing slow mobile app launches and frontend interactions across iOS, Android, and browsers.
Vercel Sandbox with Drives now starts up to 2.4x faster
Vercel Sandboxes that mount Drives now start faster. Median creation time dropped from 1.7s to 0.7s, and p95 dropped from 2.2s to 1.6s. The speedup applies automatically to every sandbox that uses Drives, with no SDK upgrade or configuration changes needed. Pricing stays the same. Learn more in the Vercel Drives documentation . Read more
Friday, October 30, 20267 releases
Data Cache is moving to Runtime Cache
Vercel is migrating Data Cache to Runtime Cache, which offers increased storage allocations and new usage-based pricing.
Amazon SageMaker AI inference now supports GPU partitioning for real-time endpoints
SageMaker AI now supports GPU partitioning using NVIDIA Multi-Instance GPU (MIG) technology. This allows multiple models to share a single GPU with hardware-enforced isolation.
AWS Glue Data Catalog now supports Apache Iceberg table optimization in the US West (N. California) Region
Automatic table optimization for Apache Iceberg is now available in the US West (N. California) region. It compacts small files and removes unreferenced snapshots to improve query performance.
Cloud Run jobs support GPU acceleration for up to 12 hours
Cloud Run jobs can now be configured with GPU support for execution durations of up to 12 hours.
BigQuery supports Apache Arrow format for query results
BigQuery now supports retrieving query results in Apache Arrow format. This preview feature improves serialization efficiency and retrieval speed.
Hybrid table support for Google Cloud (Preview)
Snowflake hybrid tables are now supported in all Google Cloud commercial regions in preview.
Introducing DBArena: open, reproducible benchmarks for hosted Postgres
DBArena provides open, reproducible benchmarks for hosted Postgres providers. It evaluates performance and pricing using a public methodology.
Thursday, October 29, 20266 releases
Real-time mode in Lakeflow pipelines is now generally available
Real-time mode in Lakeflow pipelines is generally available on classic and serverless compute, providing operational streaming processing latencies as low as five milliseconds.
Bringing Rust-native Spark compatibility to Microsoft OneLake with Sail and LakeSail
Sail, a Rust-native distributed compute framework, and LakeSail Cloud now connect to Microsoft OneLake. This integration allows teams to run Spark and Python workloads on Fabric using Delta Lake and Apache Iceberg formats without data copies or extra connectors.
Node.js 26 LTS is now Generally Available on Vercel
Node.js 26 LTS is now generally available for builds and Vercel Functions, featuring HTTP/2 support by default and the Temporal API.
Amazon SageMaker AI now supports SOCI-indexed images to start training jobs in seconds
Amazon SageMaker AI now supports Seekable OCI (SOCI)-indexed container images, allowing training and processing jobs to lazy-load images and start in seconds instead of minutes.
Wednesday, October 28, 202610 releases
Multiple real-time flows per serverless pipeline and source stage adaptive repartitioning are in Beta
Databricks released Beta support for running multiple real-time flows per serverless pipeline and adaptive source repartitioning for Kafka and Kinesis in preview channel pipelines.
Serverless GPU for clean room notebooks is now generally available
Databricks has made serverless GPU compute generally available for clean room notebooks, enabling model inference and training over shared data.
More ways to use scalar UDFs in Microsoft Fabric Data Warehouse (Preview)
Microsoft Fabric updated scalar user-defined functions (UDFs) in the Data Warehouse, adding support for procedural logic like loops and conditional branches. These functions can now be inlined for better performance across analytical workloads.
The new cost-based join optimizer with reordering support for non-inner joins
ClickHouse released a cost-based join optimizer that automatically reorders inner, outer, semi, and anti joins to determine the most efficient execution plan.
Introducing time series collections for the next generation of Amazon OpenSearch Serverless
Amazon OpenSearch Serverless now supports time series collections for ingesting and analyzing logs, metrics, and traces. This includes a lower-cost warm storage tier and built-in alerting and anomaly detection.
How we rebuilt provider routing for better, faster inference
OpenRouter updated its provider routing algorithm to continuously monitor endpoint performance, optimizing speed, cost, and reliability for inference requests.
Vercel Container Registry now supports 2 GB compressed image layers
Vercel Container Registry now supports compressed image layers up to 2 GB, facilitating the use of larger toolchains and dependency sets.
TTS output in resampled formats now starts without the extra 40 ms of silence
Gradium updated its TTS output to remove 40 ms of silence at the start of resampled audio chunks. Developers should now use start_s and stop_s parameters instead of assuming fixed chunk sizes.
See what's happening across your agents and queues
Buildkite has updated its queue pages to provide better visibility into agent capacity and job status. New charts allow teams to track agent and job activity over time to identify bottlenecks and manage capacity more effectively.
Tuesday, October 27, 20267 releases
Revamped query editor, now generally available
The revamped query editor is now generally available and is the default experience when you edit a panel. It brings your whole data pipeline into a single view (queries, expressions, transformations, and the alert rules that depend on them), so complex and multi-source panels are faster to build, review, and troubleshoot.
Build clearer, more expressive Real-Time Dashboards with new design and KPI capabilities (Generally Available)
New design and KPI features for Microsoft Fabric Real-Time Dashboards are now generally available. These include Design Tiles for layout structure, multi-KPI grouping, and enhanced KPI styling to improve dashboard readability and reduce authoring effort.
Sentry MCP: Intentional vs. unintentional drops, better field choices, and faster project-scoped requests
Agents using Sentry MCP can now tell configured drops from lost data, choose the right attribute in projects with many fields, and wait less on project-scoped connections.
Amazon ElastiCache expands data tiering to R8gd and R7gd nodes in Europe (Spain) and Asia Pacific (Hyderabad)
Amazon ElastiCache for Valkey data tiering is now available on Graviton4-based R8gd nodes in Europe (Spain) and Graviton3-based R7gd nodes in Europe (Spain) and Asia Pacific (Hyderabad). This allows for cost-effective scaling using SSDs.
Monday, October 26, 20269 releases
Gems of DuckDB 2.0.0
The DuckDB 2.0.0 Python client now includes an appender for efficient bulk data insertion. This allows users to collect rows and write them in a single operation.
Python 3.11 support is deprecated
Heroku has deprecated support for Python 3.11, with removal from buildpacks scheduled for January 2028.
Autopilot for sensitive data classification (Public Preview)
With this preview release, Autopilot is available for sensitive data classification. Autopilot lets Snowflake manage classification timing, reclassification schedules, and minimum object age to optimize performance and cost.
Secondary indexes (General availability)
Secondary indexes are now generally available in Snowflake.
Improved Lakeflow Performance Observability is now generally available
Lakeflow Jobs now provide job-level query performance metrics, insights, and optimization suggestions via Genie Code.
Databricks Runtime 19: October 26, 2026
Databricks updated Databricks Runtime 19 with operating system security updates and general bug fixes.
Databricks SQL version 2026.43 is rolling out in Current
Databricks launched Databricks SQL version 2026.43 on the Current release channel.
Faster pnpm installs
Vercel has optimized pnpm installs by reusing dependency verification across deployments, reducing build times.
Friday, October 23, 20266 releases
Atomic deployments, centralized configuration, and advanced branching for task graphs (Preview)
Snowflake has introduced a preview for task graph management, including centralized configuration, non-suspending pipeline updates, and conditional execution based on upstream outcomes.
Snowpark Container Services GEN_ARM_G1 and GEN_X64_G1 instances on Google Cloud (General availability)
Snowflake's GEN_ARM_G1 and GEN_X64_G1 instance families for Snowpark Container Services are now generally available on Google Cloud.
DataFrame caching on serverless compute is now in Public Preview
Databricks now supports DataFrame caching on serverless compute via df.cache() or df.persist(), allowing users to reuse results within a session.
citus
Citus 13.5.0 includes performance improvements for extension updates and cluster health checks, along with several bug fixes.
Deployments now show build cache size
Vercel deployment summaries now display build cache size and usage relative to limits, helping users monitor billable build time.
Thursday, October 22, 20266 releases
Apache Kafka 4.4.0 release
Apache Kafka 4.4.0 introduces new features and improvements across broker, controller, producer, consumer, admin client, and Kafka Streams components. This release includes updates to cluster metadata bootstrapping, share group dead-letter queues, and transactional state stores.
Shared Serverless (Beta)
Shared Serverless allows notebooks and files within the same bundle to share serverless compute resources and environment configurations defined in a pyproject.toml file.
ThinkingBox-Training released
ThinkingBox-Training is now available to post-train agents using executable database state as a reward. The framework uses offline policy-relative difficulty to select training tasks and improve agent dependability.
Validate GPU cluster health with Cluster Health Checks
Oracle Cloud Infrastructure introduced Cluster Health Checks to identify unhealthy GPU nodes and validate cluster readiness for AI and high-performance computing workloads.
Custom runtime images for ML Jobs (General availability)
Custom runtime images for ML Jobs are now generally available in Snowflake. Users can register custom container images to ensure reproducibility and meet specific compliance requirements.
Create purpose-built custom visuals in Real-Time Dashboards with Copilot (Preview)
Users can now use natural language to create custom HTML-based visualizations and KQL queries directly within Real-Time Dashboards using Copilot.