Releasesby GlyphNew Retirement Date for On-Demand Models in Abu Dhabi
The retirement date for the Cohere Command A Vision ( cohere.command-a-vision ) and Cohere Embed 4 ( cohere.embed-v4.0 ) on-demand models in the UAE Central (Abu Dhabi) region has changed from October 12, 2026, to October 21, 2026. After retirement, on-demand inference calls to these models in the UAE Central (Abu Dhabi) region will no longer work.
Also shipped Sep 21, 2026Oracle in September 2026 →
BigQuery Studio query results pane shows recent run history
The Query results pane now shows a short history of recent runs for a query, including multi-statement queries, without switching to the Job history tab. It is generally available.
The query history system table is now generally available
The system.query.history system table is now generally available. Use it to access account-wide records for queries run on SQL warehouses or serverless compute, including execution status, duration metrics, and query source. See Query history system table reference .
Cortex AI Function Evaluation for measuring quality (Public Preview)
Snowflake Cortex AI Function Evaluation is now available in public preview, enabling customers to measure the output quality, cost, and token usage of custom AI Functions and Cortex AI calls against labeled datasets.
Cortex AI Function Optimization for more efficient AI implementations (Public Preview)
Snowflake Cortex AI Function Optimization is now available in public preview, enabling customers to automatically search across prompts and models for improved implementations of custom AI Functions.
Cortex Agents Compact API (Preview)
The Cortex Agents Compact API is now available in preview. Use the agent:compact endpoint to summarize a conversation and pass the compact representation to subsequent agent:run requests. Compacting conversation history reduces token consumption and helps conversations stay within the model context window.
You can now use the AlloyDB Columnar Engine as a read-optimized, in-memory cache for HNSW vector indexes
You can now use the AlloyDB Columnar Engine as a read-optimized, in-memory cache for HNSW vector indexes. This feature is generally available ( GA ). It accelerates vector search performance and increases queries per second (QPS) for vector workloads. For more information, see Accelerate queries with the Columnar Engine .
More from Oracle Cloud InfrastructureEvery Oracle Cloud Infrastructure release →
Share and Run the Select AI Agent Framework teams Across Schemas
You can now share the Select AI Agent Framework teams across database schemas. Define and manage an agent team in one schema, then grant another database user or role permission to run the team by using an owner-qualified team name.
Select AI for Java
Select AI for Java provides a Java SDK for using Select AI capabilities in Java applications. Use the Java-native API to work with AI profiles, credentials, conversations, vector indexes, natural-language SQL, RAG, summarization, translation, feedback, and synthetic data generation. See Select AI for Java for more information.
Bring Your Own Reservations for Data Science
You can use existing Oracle Cloud Infrastructure (OCI) Compute capacity reservations with supported Data Science resources, including notebook sessions, single-node jobs and job runs, and model deployments. Bring Your Own Reservations (BYOR) uses Compute capacity that's reserved in your tenancy without transferring the reservation to Data Science.
Route on-demand inference requests across regions with smart model router
OCI Generative AI now supports smart model router for cross-region routing of on-demand inference requests within a user-selected regional scope. Create a routing profile that specifies an on-demand model and select regions within the same realm where that model is available. Then reference the routing profile in a Chat API or Responses API request.
Import DeepSeek V4.1 Flash into OCI Generative AI
You can now import the deepseek-ai/DeepSeek-V4.1-Flash model into OCI Generative AI, create endpoints for it, and use it in the Generative AI service. The model supports text, image, and combined image and text inputs and generates text.
Import Google Gemma 4 26B A4B IT into OCI Generative AI
You can now import the google/gemma-4-26B-A4B-it model into OCI Generative AI, create endpoints for it, and use it in the Generative AI service. The model supports text, image, and combined image and text inputs and generates text.
Find models by region with model discovery in OCI Generative AI
OCI Generative AI now supports model discovery. Use model discovery to find models in a specified region and review details such as: Model capability including chat, text embedding, and text reranking. Input and output types, such as IMAGE_TEXT_TO_TEXT. Whether a model supports on-demand or dedicated serving.
HeatWave: Automatic Storage Expansion Enhanced
Automatic storage expansion now increases the DB system storage when the remaining available storage is 10% or less of the currently allocated storage. The new storage size is calculated based on the current allocation, and the expanded size cannot exceed the configured maximum storage size.