This month’s Oracle Cloud Infrastructure (OCI) AI updates highlight new ways for enterprises to build more capable, context-aware AI experiences on OCI. From improved retrieval with Cohere Rerank 4 and expanded multimodal support across text, image, and voice, to broader model availability and new regional access in UAE Central (Abu Dhabi), the June updates continue to deepen OCI’s AI foundation. This edition also features new resources and cloud intelligence capabilities that help teams connect AI innovation to real-world operations, from enterprise chat architectures to resource analytics and hands-on deployment demos.
New updates and features:
Cohere Rerank 4 is now available in OCI Enterprise AI on demand and on Dedicated AI Clusters
Rerankers play an important role in enterprise AI search by refining initial retrieval results and surfacing the most relevant documents, passages, or product listings for a user’s intent.
With Cohere Rerank 4 in OCI Enterprise AI, teams can build more accurate search and agentic AI experiences with managed access to leading models, enterprise governance, and deployment options built for production AI.
Useful for teams working on RAG, knowledge search, customer support, recommendations, and other retrieval-heavy AI use cases.
Learn more about Cohere Rerank on OCI
Multimodal capabilities for OCI Enterprise AI Models
OCI Enterprise AI expanded multimodal support with new features for Cohere Embed 4, enabling embeddings from text, images, or combined text-image inputs, and xAI Voice, which converts text into spoken audio.
These updates give customers additional tools to build richer AI experiences, from visual search and document understanding to voice-enabled assistants and interactive, low-latency applications.
Learn more about new Cohere Embed features on OCI
Learn more about xAI Voice on OCI
OCI Enterprise AI expands to UAE Central (Abu Dhabi)
OCI Enterprise AI is now available in the UAE Central (Abu Dhabi) region, giving customers more regional flexibility for deploying generative AI workloads on OCI. Available models in the region may be offered on-demand or through dedicated AI clusters.
With this update, UAE-based and regional customers can better address data residency, latency, and deployment needs while building AI applications closer to their users and operations.
NVIDIA Nemotron 3 Ultra is now available in OCI Enterprise AI on Dedicated AI Clusters
NVIDIA Nemotron models are open models that bring frontier reasoning and agentic performance with fully open weights, training data, and recipes. Oracle is glad to now list Nemotron 3 Ultra model for OCI Enterprise AI imported model deployment.
With an OCI dedicated AI cluster, teams can host Nemotron 3 Ultra on Oracle-recommended GPU infrastructure and run it through a managed OCI endpoint, giving production AI workloads a more controlled deployment path.
Additional new Model Import compatible models
OCI Enterprise AI added new compatible models for import, including Alibaba Qwen and Google Gemma models, and now supports hosting OpenAI gpt-oss-20b and gpt-oss-120b on B200 dedicated AI clusters in UAE Central (Abu Dhabi).
These new models and deployment region give customers more flexibility to choose the right model and deployment architecture for their use case, with dedicated hardware options that help meet performance, capacity, and regional deployment needs.
OCI Resource Analytics for near-real-time cloud estate intelligence
OCI Resource Analytics helps teams understand OCI resources, relationships, and configuration metadata across regions and tenancies, with access through SQL queries, graph-based visualizations, and analytics dashboards.
Because Resource Analytics runs on Oracle’s AI Database foundation, teams can also take advantage of database-native AI capabilities such as select AI and MCP server support, helping make cloud estate data more accessible to AI assistants, agents, and automation workflows.
For customers managing fast-growing cloud environments, Resource Analytics goes beyond basic resource search to show how resources connect and where dependencies may affect operations. Teams can use that context to reduce operational risk, give automation more reliable current-state data, and identify where hidden waste may be growing.
New resources for AI at OCI:
Live Demo Day: Oracle and NVIDIA Accelerate AI Deployments on OCI
Join Oracle and NVIDIA experts for a demo-first webinar that shows exactly how to deploy OCI AI Accelerator Packs, Oracle’s preconfigured, self-service AI solutions that you can launch quickly and easily from within the OCI Console.
See how Oracle Cloud Infrastructure and NVIDIA are partnering to deliver enterprise-grade AI solutions, built for business outcomes.
New GitHub Asset: OCI Enterprise AI Chat Reference Architecture and Deployment Guide
A new GitHub asset, OCI Enterprise AI Chat, provides a reference architecture and deployment guidance for building secure, enterprise-grade AI chat applications with OCI Enterprise AI.
This resource helps organizations accelerate generative AI adoption with proven implementation patterns, reducing development effort and enabling faster deployment of production-ready AI assistants that meet enterprise security and governance requirements.
New Video: OCI Enterprise AI: Flexible Model Routing on Your Terms
See how OCI Enterprise AI helps organizations build more flexible, cost-efficient AI strategies by making it easier to work across multiple models instead of relying on a one-size-fits-all approach
Model choice helps customers optimize for performance, cost, and use case needs, making enterprise AI deployments more adaptable and resilient.
For more Oracle AI News: What’s New in Oracle AI? May 2026 edition