Releasesby GlyphNew H100 Unit Shape for OpenAI gpt-oss-120b
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.
Also shipped Oct 30, 2026Oracle in October 2026 →
Microsoft Graph Logs connector is available in Beta
The Microsoft Graph Logs connector is now in beta for Databricks Lakeflow Connect, supporting incremental ingestion of security alerts from Microsoft security products.
Hybrid table support for Google Cloud (Preview)
Snowflake hybrid tables are now supported in all Google Cloud commercial regions in preview.
External OAuth for Snowflake Postgres (General availability)
External OAuth for Snowflake Postgres is now generally available, allowing authentication via external identity providers.
Spanner supports equality and ordering comparisons on JSONB and JSON values
Spanner now allows equality and ordering comparisons on JSONB and JSON values directly in GROUP_BY and ORDER BY clauses. It also supports implicit type coercion and search indexes for JSON equality operations.
[In preview] Public Preview: Support database-level managed identity (including FIC) in Create Database Scoped Credential T-SQL
Users can now assign a user-assigned managed identity to individual databases in Azure SQL Database. This enables secure access to files in Azure Blob Storage and Data Lake Storage Gen2 via external tables and OPENROWSET.
Inference request metadata logging (GA)
Inference request metadata logging is now generally available for Gemini and Claude models. It records model usage, latency, and error codes without capturing prompt or response content.
More from Oracle Cloud InfrastructureEvery Oracle Cloud Infrastructure release →
New tutorial: Search Documents with Vector Stores and OCI Generative AI API Keys
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.
New tutorial: Discover Models with the Python SDK and REST API
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.
On-demand model access in Jeddah, Jerusalem, and Marseille
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.
Remove Email Suppressions with the UserEmailSuppressionRemover API
The new UserEmailSuppressionRemover API allows users to remove email suppressions. This enables the receipt of Identity Domains email notifications.
Oracle Cost Advisor for AI in Limited Availability
Oracle Cost Advisor for AI is now in limited availability for OCI Generative AI. It provides a unified dashboard to track AI spending across OCI, OpenAI, and Anthropic, and offers recommendations for cost-optimized model usage.
Manage conversation context with compaction in the OCI Responses API
The OCI Responses API now supports manual and automatic context compaction to manage conversation history in long-running agent workflows. This feature is available for all models supported by the API.
Client ID Metadata Document (CIMD) Support for Database Tools MCP Server
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.
Import NVIDIA Nemotron 3 Super and Ultra in US Government Cloud and US Defense Cloud
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.