This month’s Oracle AI updates focus on giving organizations more flexibility in how they build, deploy, and manage enterprise AI on OCI. Highlights include expanded model and infrastructure options in OCI Enterprise AI, new capabilities that strengthen security and scalability, and tools that simplify working with enterprise data. We also feature new educational resources on AI cost optimization and model routing, along with registration for Oracle AI World 2026, where you can explore the latest AI innovations firsthand.
New updates and features:
OCI Enterprise AI Hosted Application Endpoints Now Support OCI IAM Authentication
Hosted application endpoints in OCI Enterprise AI now support OCI Identity and Access Management (IAM) authentication, allowing organizations to manage access using the same identity and policy framework they already use across OCI. This simplifies user and application authentication, helps strengthen security, and makes it easier to integrate hosted AI applications into existing enterprise governance workflows.
H100 Multi-Node Serving Now Available for Imported Models
OCI Enterprise AI now supports H100 multi-node serving for imported models, enabling customers to deploy large models across multiple H100 GPU nodes. This allows organizations to serve larger, more demanding models with additional scalability and performance while reducing the operational complexity of managing distributed AI infrastructure.
Expanded On-Demand Model Availability in OCI Enterprise AI
OCI Enterprise AI has expanded its portfolio of models available through on-demand inference, giving customers more choices without requiring dedicated infrastructure. With additional models available on demand, organizations can more easily evaluate new models, optimize cost and performance, and scale AI workloads based on changing business needs.
See the full list of available models
More Hardware Choice for Cohere and Meta Models
OCI Enterprise AI now offers additional hardware unit shapes for Cohere and Meta models, giving customers greater visibility into the infrastructure powering their AI workloads and the flexibility to choose the hardware that best meets their performance and cost requirements. This helps deliver more predictable application performance while helping organizations align infrastructure selection with workload demands.
Learn more about hardware unit shapes
Background Mode Now Available for the OCI Responses API
The OCI Responses API now supports Background Mode, allowing long-running AI tasks to execute asynchronously without requiring applications to maintain an active connection. This makes it easier to build reliable AI workflows for complex reasoning, document processing, and agentic tasks that may take longer to complete while improving application responsiveness.
Natural-Language Access to Oracle Database Data
Natural-language access to data is now available directly in the Oracle Database Console and through the OCI Enterprise AI SQL Assistant MCP Toolset, allowing users to query enterprise data using everyday language. By translating natural-language requests into SQL, these capabilities make it easier for developers, analysts, and business users to explore data while helping accelerate AI-powered application development.
Learn more about natural language to SQL
OCI Enterprise AI is now available in dedicated cloud regions
In July, we announced OCI Enterprise AI for OCI Dedicated Cloud, enabling organizations to adopt production AI while keeping sensitive data and workloads within required security, governance, residency, and sovereignty boundaries.
Customers can use Oracle-managed models, import validated third-party or open-source models and offer proprietary models through a consistent OCI interface.
Learn more
New resources for AI at OCI:
What Is LLM Routing?
The What Is LLM Routing? article explains how routing requests across multiple large language models can help improve AI application quality, performance, and cost efficiency. It explores routing strategies, common use cases, and best practices for matching each request to the model best suited for the task, helping organizations build more efficient multi-model AI architectures.
What Is AI Token Economics?
Oracle’s new What Is AI Token Economics? article explores how token usage directly impacts AI application costs and performance, and why understanding token consumption is critical for scaling enterprise AI. It outlines practical strategies for managing token usage, helping organizations optimize AI spending while maintaining application quality and responsiveness.
Learn more about AI Token Economics
Join us at Oracle AI World 2026
Registration is now open for Oracle AI World 2026, Oracle’s flagship AI event bringing together customers, partners, developers, and AI experts to explore the latest innovations in AI. Attendees can discover new product announcements, hear customer success stories, participate in technical sessions and hands-on learning, and connect with Oracle AI leaders.
For more Oracle AI news: What’s New in Oracle AI? July 2026 Edition