Azure OpenAI

Azure OpenAI releases gpt-image-1-mini model for global deployments

Azure OpenAI has released gpt-image-1-mini, a smaller and more cost-effective version of the gpt-image-1 model for image generation. The model supports most use cases except image edits and input fidelity features. Users need to request limited access and can follow the image generation guide to get started.

Date not specified

Azure OpenAI adds built-in PII detection content filter

A new content filter for detecting personally identifiable information (PII) is now available as a built-in feature. This enhancement allows users to automatically identify and block sensitive information in LLM outputs to improve data privacy protection. No immediate action is required, but users can review the documentation to implement this optional security feature.

Date not specified

OpenAI releases gpt-5-codex model with CLI and VS Code support

Azure OpenAI has introduced gpt-5-codex, a new reasoning model designed for code development workflows. The model requires registration for access and integrates with Codex CLI and Visual Studio Code extension. Users with existing access to gpt-5 will automatically receive access to this new model.

Date not specified

OpenAI Sora adds video-to-video generation capability

The Sora model now supports video-to-video generation, allowing users to provide a short video as input to generate longer videos that incorporate the original content. This expands Sora's capabilities beyond text-to-video generation. No action required - this is an additional feature enhancement.

Date not specified

Azure OpenAI Spillover Feature Now Generally Available

Azure OpenAI has released Spillover as a GA feature that automatically routes traffic overages from provisioned deployments to standard deployments during traffic fluctuations. This helps maximize utilization of provisioned capacity by providing overflow handling. No immediate action required, but users can optionally configure spillover to optimize their deployment efficiency.

Date not specified

Azure AI Foundry adds GPT-5 series support via Model Router

The Model Router for Azure AI Foundry now supports GPT-5 series models, automatically selecting the best underlying chat model for each prompt. This enhancement expands the available model options within the existing router functionality. No immediate action required, but users can leverage the new GPT-5 capabilities through the existing Completions API.

Date not specified

Azure OpenAI Sora model adds image-to-video generation capability

The Sora model now supports converting images to videos with flexible frame positioning options. The enhanced model is available in Sweden Central and East US 2 regions. No action required - this is an additive feature enhancement.

Date not specified

OpenAI GPT RealTime & Audio Models GA on Azure AI Foundry

Azure AI Foundry now offers general availability of OpenAI's GPT RealTime and Audio models with enhanced instruction following, new voices, improved audio quality, and real-time API improvements including image input and conversation mode. Customers are recommended to transition to the new GA models to access the latest features. No immediate action required but migration is advised for optimal functionality.

Date not specified

OpenAI Releases GPT-5 Model Family (gpt-5, gpt-5-mini, gpt-5-nano, gpt-5-chat)

Azure OpenAI has released the new GPT-5 model family with four variants. The base gpt-5 model requires registration and supports Provisioned Throughput Units, while gpt-5-mini, gpt-5-nano, and gpt-5-chat are available without registration requirements. This expands model options for developers with different performance and access needs.

Date not specified

Azure OpenAI adds input fidelity and partial image streaming to image APIs

New input_fidelity parameter allows better control over style preservation in image edits, useful for maintaining facial features and brand identity. Partial image streaming provides real-time visual feedback during image generation. These are additive features that enhance existing image generation capabilities without breaking changes.

Date not specified

Azure OpenAI releases codex-miniando3-proare reasoning model

Azure OpenAI has released a new reasoning model called codex-miniando3-proare. This appears to be a new capability addition to their model offerings. Users can refer to the getting started documentation to learn about implementation and usage.

Date not specified

Azure OpenAI introduces Spotlighting sub-feature for enhanced prompt shield protection

Azure OpenAI has released Spotlighting, a new sub-feature of prompt shields that provides enhanced protection against indirect attacks from embedded documents. The feature uses special formatting to tag input documents as lower trust to the model. This is an informational release of new security functionality with no immediate action required.

Date not specified

Azure OpenAI launches Sora video generation model

OpenAI's Sora model is now available on Azure OpenAI, enabling users to generate realistic and imaginative video scenes from text instructions. This new capability expands Azure's AI offerings into video generation. Users can get started with the Video generation quickstart guide.

Date not specified

Azure AI Foundry launches Model Router for automatic chat model selection

Azure AI Foundry has released a new deployable Model Router that automatically selects the optimal underlying chat model for each prompt. This new capability enhances response quality by intelligently routing requests to the most suitable model. No action required - this is an optional new feature available through the Completions API.

Date not specified

Azure OpenAI releases GPT 4.1 and GPT 4.1-nano with 1M token context

Azure OpenAI has released new GPT 4.1 and GPT 4.1-nano models, with GPT 4.1 offering an expanded 1 million token context window. This represents a significant increase in context capacity for applications requiring processing of large documents or extended conversations. No immediate action required, but users may want to evaluate these new models for their use cases.

Date not specified

Azure OpenAI releases o1-mini and o1 reasoning models with enhanced performance

Azure OpenAI has launched new o1-mini and o1 reasoning models that offer improved reasoning capabilities, quality, and performance compared to previous models. This is a new feature release that expands available model options for users. No immediate action is required, but users may want to explore these new models for enhanced reasoning tasks.

Date not specified

OpenAI releases GPT-4o audio models for transcription and text-to-speech

Azure OpenAI has released new audio models including gpt-4o-transcribe, gpt-4o-mini-transcribe for speech-to-text, and gpt-4o-mini-tts for text-to-speech. These models are accessible through the /audio and /realtime APIs, expanding Azure's audio processing capabilities. No action required, but developers can now leverage these new models for enhanced audio functionality.

Date not specified

Azure OpenAI Realtime API adds WebRTC support for low-latency audio streaming

The Realtime API preview now includes WebRTC functionality, enabling real-time audio streaming for applications like live customer support and interactive voice assistants. This enhancement provides lower latency interactions but remains in preview status. No immediate action required for existing implementations.

Date not specified

Azure OpenAI releases GPT-image-1 model with enhanced image generation and editing

Azure OpenAI has launched GPT-image-1, a new image generation model that improves upon DALL-E with better instruction following, text rendering, and new image editing/inpainting features through image input support. The model requires application for limited access and follows existing image generation implementation patterns.

Date not specified

Azure OpenAI launches Responses API with computer-use-preview model

Azure OpenAI introduces a new stateful Responses API combining chat completions and assistants capabilities, plus a computer-use-preview model requiring special access approval. This is a new feature addition that enhances existing capabilities without breaking changes. Users interested in computer-use functionality need to apply for limited access through Microsoft's eligibility process.

Date not specified