Container Registry now supports pull-through cache repositories and the top-level repository capability. A pull-through cache repository retrieves requested container images from an approved Container Registry endpoint in another OCI region within the same OCI realm. On the first request, Container Registry streams the image to the client while caching it in the local OCI region. Subsequent reques…
You can use Oracle Application Express (APEX) 26.1 to develop applications for Autonomous AI Database on Dedicated Exadata Infrastructure. Newly created Autonomous Container Databases (ACDs) are provisioned with APEX 26.1. Existing ACDs continue to use the APEX version with which they were provisioned. To use the latest APEX features with existing applications, submit a Service Request (SR) throug…
On Autonomous AI Database on Dedicated Exadata Infrastructure, unsupported role and privilege statements are handled automatically during migration: statements that cannot be applied are ignored, and certain roles are mapped to supported equivalents, such as DBA to PDB_DBA. YOu can review the original and converted statements and the action taken in the DBA_CONVERTED_STATEMENTS view and use the DB…
Database Tools MCP Server now enables users to manage MCP client access using Client ID Metadata Documents (CIMDs) in IAM Identity Domains. A Client ID Metadata Document enables an OAuth client to use the HTTPS URL of a client-hosted metadata document as its client_id. When CIMD is used, users only need the MCP server URL to configure their MCP clients. They no longer need to specify a client_id.…
You can now host OpenAI gpt-oss-120b (openai.gpt-oss-120b) on dedicated AI clusters using the OAI_H100_X4 unit shape in OCI Generative AI. This shape is available in all 11 OC1 commercial regions with local model processing. To use this shape, create a dedicated AI cluster with OAI_H100_X4, then create an endpoint for the model on that cluster. If your tenancy has insufficient capacity under the d…
On-demand model access through routing profiles is now available in OCI Generative AI in three more OC1 commercial regions: Saudi Arabia West (Jeddah), JED Israel Central (Jerusalem), MTZ France South (Marseille), MRS These regions have no local model processing. Create a routing profile in your region, select a supported model, and choose one or more target OCI regions where the model is availabl…
A new tutorial for OCI Generative AI shows how to explore the model catalog with the Python SDK and REST API. Learn step by step how to authenticate with OCI IAM request signing, filter models by region, capability, serving mode, and lifecycle status, and inspect API compatibility and configuration parameters. The tutorial includes ten practical Python examples for searching model names, finding c…
A new tutorial for OCI Generative AI shows how to search documents with vector stores using an OCI Generative AI API key. Follow either curl requests or OpenAI Python SDK examples to create a store, upload a sample document, ingest it, and retrieve relevant passages with a natural-language question. The tutorial also covers checking ingestion status, listing and renaming stores, troubleshooting re…
You can now use the UserEmailSuppressionRemover API to remove email suppressions associated with a user's email addresses. Removing the suppression allows the user to receive Identity Domains email notifications again. For more information, see Removing Email Suppressions for a User
The OCI Responses API now supports explicit and automatic context compaction for long-running conversations and agent workflows. Compaction reduces accumulated conversation history to a smaller representation that you can use in later requests. Call POST /responses/compact to control when an application compacts history, or enable automatic compaction on POST /responses with context_management and…
Bring your organization's AI spending into focus and turn insights into smarter spending decisions with Oracle Cost Advisor for AI. Now available in Limited Availability (LA) in OCI Generative AI in US East (Ashburn), Cost Advisor helps finance, technology, and engineering teams understand AI costs and explore opportunities to make their investments go further. See costs from all your connected AI…
You can now import the following NVIDIA Nemotron 3 models into OCI Generative AI in US Gov West (Phoenix) (OC2) and US DoD West (Phoenix) (OC3), host them on dedicated AI clusters, and create endpoints for inference: nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16 nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-NVFP4 Both models support the TEXT_TO_TEXT capability. The minimum dedicated AI cluster unit sha…
IAM identity domains now prevents administrators, including Identity Domain Administrators, from updating, replacing, or deleting Oracle-seeded application roles (AppRoles). This protection helps prevent accidental or unsupported changes to roles required for identity domain functionality. No customer action is required. Custom AppRoles and customer-created applications are unaffected. To access O…
Oracle Cloud Infrastructure introduces Cluster Health Checks (CHC) to help you identify unhealthy GPU nodes and validate cluster readiness before running AI and high-performance computing workloads. CHC provides two kinds of health checks: Passive checks (through Oracle Cloud Agent) evaluate GPU, driver, and network health on supported instances. Optional active tests (through Slurm or Kubernetes)…
You can now configure OKE to coordinate maintenance event repair for eligible bare metal GPU worker nodes in enhanced clusters. When the GPU Health Monitoring add-on reports an eligible critical HPC fault, OKE correlates the fault with an OCI Compute maintenance event. For nodes selected for repair, OKE makes the node unavailable for new workloads and attempts to drain workloads from the node befo…
You can now select No authentication when creating a hosted application in OCI Generative AI. Use this option to serve public content or run applications that manage their own authentication. Anyone with access to the application endpoint can send requests without platform-managed authentication. To restrict access to applications, implement authentication and authorization in your application. OC…
Hosted applications in OCI Generative AI now support persistent file storage. Store model artifacts, intermediate results, checkpoints, and application state at a configured container path across restarts and redeployments. Use the Console or API to add up to 50 GB of file storage when creating a hosted application. This release supports one application replica with file storage. Set the minimum a…
This Oracle Cloud Infrastructure (OCI) Device Data FHIR Service release includes these capabilities: Create and restore resources with PUT: Use client-assigned logical IDs to create resources, update existing resources, and restore logically deleted resources in FHIR R4 and R6 Ballot 4. Creation and restoration require both create and update permission and return 201 Created; updating a live resou…
Full Stack Disaster Recovery introduces the following new features. Now, you can: Add a PostgreSQL DB System to a Disaster Recovery (DR) Protection Group and orchestrate switchovers and failovers between primary and warm standby DB Systems across OCI regions. See the following new topics: Preparing PostgreSQL DB System for Disaster Recovery Add a PostgreSQL DB System to a Disaster Recovery Protect…
Use SQL tracing to identify the source of excessive database workloads, such as high-load SQL statements in your application. You can trace the current session with ALTER SESSION, or use the DBMS_USERDIAG package to target a specific SQL statement, session, client identifier, or service, module, and action. You can also collect bind variable, wait event, and execution plan information for more det…
HeatWave now supports MySQL database versions 26.10.0, 9.7.4, and 8.4.13. These versions provide security and bug fixes. New DB systems are now based on version 9.7.4 by default. You can optionally select version 26.10.0 or 8.4.13 when creating a new DB system. For existing DB systems, you can manually upgrade them to a newer version, such as from version 8.4.11 to 8.4.13, or allow your maintenanc…
OCI Generative AI now supports access to on-demand models through routing profiles in 30 OC1 commercial regions. Call supported models processed in other regions from your region’s playground or through the Chat API, OCI Chat Completions API, or OCI Responses API, when the selected model supports the API. These regions also support selected Enterprise AI Agents and Enterprise AI Governance feature…
GoldenGate adds support for the following AI Services and Processing Engines connection types: OCI Generative AI Open AI Google Gemini Voyage AI Apache Spark Databricks Engine Oracle AI Data Platform Engine Learn more about what's new in this release.
OCI Database with PostgreSQL now supports in-place major version upgrades, allowing you to upgrade an existing database system to a newer PostgreSQL major version. This capability helps you keep your database platform current and take advantage of features and improvements available in newer PostgreSQL releases. For more information, see pg_upgrade.
You can now import the nvidia/NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-NVFP4 model into OCI Generative AI, create endpoints for it, and use it for image and text understanding. In OCI Generative AI, use the following details for this model: Model capability: IMAGE_TEXT_TO_TEXT Minimum dedicated AI cluster unit shapes: H100_X2 and H200_X2 For the complete list of models that you can import, see Model…
You can now import the Qwen/Qwen3.6-35B-A3B-FP8 model into OCI Generative AI, create endpoints for it, and use it for image and text understanding. In OCI Generative AI, use the following details for this model: Model capability: IMAGE_TEXT_TO_TEXT Minimum dedicated AI cluster unit shapes: A100_40G_X2, A100_80G_X1, H100_X1, and H200_X1 For the complete list of models that you can import, see Model…
OCI Generative AI now supports CPU-only access to on-demand models through routing profiles in 30 OC1 commercial regions. Call supported models hosted in other regions from your region’s playground or through the API. These regions also support selected Enterprise AI Agents and Enterprise AI Governance features, including hosted deployments, projects, API keys, and the OCI Responses API. For regio…
Sandboxes are now available in OCI Generative AI. A sandbox provides a managed, temporary Linux workspace where developers and AI agents can run code and shell commands. Each sandbox has its own file system, network namespace, and process space. Use the OCI CLI or SDK to: Create sandboxes within a Generative AI project. Select Compute resources, set the sandbox lifetime, and configure command time…
Oracle Data Safe has a new Crypto Posture Assessment feature that provides a centralized view of the cryptographic posture of your registered target databases. For more information see Crypto Posture Assessment Overview in the Oracle Data Safe guide.
Java Management supports the following releases that are part of the Oracle Critical Patch Update (CPU) August 2026 for Oracle Java SE: 27.0.1: https://www.oracle.com/java/technologies/javase/27-0-1-relnotes.html 25.0.5: https://www.oracle.com/java/technologies/javase/25-0-5-relnotes.html 21.0.13: https://www.oracle.com/java/technologies/javase/21-0-13-relnotes.html 17.0.21: https://www.oracle.com…