Introducing o1: OpenAI's new reasoning model suite for developers and enterprises on Azure

We are excited to add OpenAI’s latest o1-preview and o1-mini models to Azure OpenAI Service, Azure AI Studio, and GitHub Models.

We are excited to add OpenAI's latest o1-preview and o1-mini models to Microsoft Azure OpenAI Service, Azure AI Studio, and GitHub Models. The o1 series enables complex coding, mathematical reasoning, brainstorming, and comparative analysis capabilities, setting a new benchmark for AI-powered solutions.

The o1-preview and o1-mini models are now accessible in Azure AI Studio and GitHub Models for a select group of Azure customers to collaboratively explore and identify the unique strengths of each model. The o1 series of advanced reasoning models are well suited to complex and nuanced problem areas such as these:

• Complex code generation: Can generate algorithms and perform advanced coding tasks to assist developers.

• Advanced troubleshooting: Perfect for comprehensive brainstorming sessions and tackling complex problems.

• Complex document comparison: Ideal for analyzing contracts, case files or legal documents to identify subtle differences.

• Following instructions and workflow management: Particularly suitable for handling workflows that require a shorter context.

Early adopters share their insights

Our early adopters have already shared valuable insights and are excited to leverage these models to enhance the capabilities of their applications.

GitHub Copilotthe world's most widely used AI developer tool, has tested the O1 models and shown promising results in code analysis and optimization.

“With o1 and its strong reasoning capabilities, GitHub Copilot enables developers to develop the big picture faster. Nothing beats the feeling of solving a coding problem in minutes instead of hours. And with GitHub Models, we can't wait to see what developers do with o1 in their apps.” – Thomas Dohmke, CEO, GitHub.

Harveya leader in generative AI for professional services, is delighting customers by integrating its platform with the Azure OpenAI service, now available in the Azure Marketplace.

“Based on our testing, we believe o1 represents a game-changer in legal reasoning, allowing us to build customized solutions that handle more sophisticated workflows such as SPA and S-1 drafts or agents assisting with critical aspects of due diligence and e-discovery. This marks a transition from chatbots to collaborative platforms that enable professionals to work hand-in-hand with AI on complex use cases.” – Winston Weinberg, co-founder and CEO of Harvey.

knowledgean applied AI lab that builds end-to-end software agents, is excited to integrate the new models into Devin, the world's first fully autonomous AI software developer.

“o1 represents a significant advancement in inference models, and we are excited to see how innovations like this will improve Devin and enable it to solve increasingly complex coding tasks.” – Scott Wu, CEO of Cognition.

These examples illustrate just some of the ways the O1 Series models can bring advanced thinking and analysis capabilities to your projects.

New updates on our commitment to responsible AI and safety as standard

As we deploy new generative AI models, we continue to prioritize security. The o1 and o1-mini models in the Azure OpenAI Service have content safety features enabled by default at no additional cost to you.

OpenAI has invested in additional security improvements to these models, including new methods to help the model reject unsafe requests. This makes the o1 series of models the most robust models we have deployed to date. You can assess the security of your AI application and model deployments with Azure AI Studio security assessments.

As part of our joint efforts with OpenAI to make models more secure, we continue to expand and improve the content safety systems across all models. We're excited to announce that Prompt Shields and Protected Materials for Text are now generally available in Azure OpenAI Service. These features help customers maintain the security and integrity of their generative AI deployments and prevent potential misuse. Finally, we introduced Spotlighting, a family of techniques designed to help models distinguish between valid instructions and potentially untrustworthy external input. We've updated our documentation to explain how to use Spotlighting and related techniques so customers can benefit from the latest research from Microsoft.

Trying out the new models in Azure AI Studio and GitHub

We're excited to offer select customers a preview of the o1-preview and o1-mini models in the Azure AI Studio Playground and GitHub Models. This exclusive access allows early experimentation with their advanced reasoning capabilities. API access will be available soon. To get started, visit the Azure AI Studio Models Catalog and request access today.

Navigate through our model range

GPT-4o models have industry-leading capabilities in text and image understanding and GPT-4o mini provides high-quality summaries and Q&As with cost-effective deployment options and fast response times. At the same time, we recognized the limitations of these models when it comes to solving even simple math problems with good guidance and examples.

To better understand the place of o1-preview and o1-mini in our model portfolio, here is a quick overview of the key models that power the Azure OpenAI Service:

• o1 preview: Focuses on advanced reasoning and complex problem solving, including mathematical and scientific tasks. Ideal for applications that require deep contextual understanding and agent-based workflows.

• o1-mini: Smaller and faster and 80% cheaper than o1-preview, good performance in code generation and small context operations.

• GPT-4o: A versatile, multimodal model that excels in both text and image processing, offering superior performance in languages ​​other than English and in image processing tasks. Suitable for applications that require higher accuracy and multilingual capabilities. The model also provides JSON-structured outputs for consistent, well-defined data formats, reducing the need for post-processing and improving application efficiency. Designed for real-time applications that require fast, reliable text responses at minimal cost.

• GPT-4o Mini: A smaller, low-cost version of GPT-4o, optimized for environments with limited resources or high cost constraints. Retains text and image processing capabilities, making it ideal for lightweight applications.

• DALL-E: Safely generates images from text prompts, ideal for creative content and marketing.

• Whisper: Transcribes and translates speech to text, suitable for real-time transcription and multilingual communication.

These models represent a range of capabilities, from efficient word processing to advanced reasoning and multimodal functionality. Each model has paved the way for the innovations we continue to introduce with o1.

Innovate at scale with Azure

The latest OpenAI models are supported by all the features of the Azure platform, including enterprise-grade security, flexible deployment options, and broad regional availability that helps customers meet their data residency and compliance requirements. We have over 60,000 Azure AI customers with exciting production use cases across all industries.

Shape the future of AI with us

We're excited to invite a select group of customers to explore o1-preview and o1-mini in the Azure AI Studio Playground and experience their unique capabilities first-hand by visiting the model catalog and requesting access. But don't worry if you're not in the first group – our upcoming API release will make the o1 series models more widely available.

We believe in the power of collaboration and innovation and want everyone to have the opportunity to benefit from these groundbreaking advances in AI. Whether you're new to Azure AI or an experienced user, there's never been a better time to experiment, innovate, and grow with us. We look forward to welcoming a wider audience soon and shaping the future of AI together.