Announcing the o1 model in Azure OpenAI

We are excited to announce that the o1 model will soon be available on Microsoft Azure OpenAI. This multimodal model brings advanced reasoning capabilities and enhancements that will significantly improve your AI applications and solutions. The o1 model supports both text and visual inputs, making it ideal for a wide range of applications from complex problem solving to contextual AI.

Customers using the o1 model in Azure OpenAI

Rohirrim: o1 model analysis “awesome”

Rohirrim Inc. is a fast-growing AI startup that built Rohan, a secure, knowledge-driven generative AI platform designed for the Fortune 1000 and aerospace and defense sectors. Rohan streamlines complex RFP responses, eliminates the need to sift through unstructured data, and enables quick and accurate responses with a company’s unique voice. Hosted as a dedicated platform as a service (PaaS), it ensures data control, high compliance and low management risk, increasing productivity and reducing burnout.

With o1, we’ve unlocked a layer of contextual understanding and fine-grained analysis that is, frankly, mind-blowing. These developments mean we can now apply AI to the complex, detail-intensive aspects of design management, while freeing up human experts to focus on higher-order intuitive thinking – exactly the kind of insight and creativity that really helps organizations grow. —Steven Aberle, CEO, Rohirrim

Wrtn Technologies: Response Accuracy “Increased Exponentially”

Wrtn Technologies is an AI-based solutions provider recognized for its innovative approach and commitment to excellence. Leveraging the advanced capabilities of the o1 model, Wrtn Technologies has revolutionized its digital presence and performance.

With the o1 model, our success rate for accurate answers to questions has increased exponentially according to our measurements.—Dongjae “DJ” Lee, Chief Product Officer, Wrtn Technologies

Harvey: o1 is a “sophisticated thinker”

Harvey is a generative AI platform trusted by law firms and professional service providers. It helps lawyers and other professionals by automating complex legal tasks such as summarizing and comparing documents, performing due diligence and referencing case law. By leveraging Azure AI infrastructure, Harvey increases efficiency and productivity in the legal industry.

What you have is an incredibly powerful model, a sophisticated thinker. Can create plans from scratch to solve complex tasks. And there is also a degree of controllability through following the guidelines. … (You can) put a human in the loop and really steer the model toward higher quality outputs.—Niko Grupen, Head of Applied Research, Harvey

See o1 in action as Harvey uses o1 to analyze a range of complex legal documents here.

New o1 features for developers

The o1 model introduces several new features that can enhance your reasoning-based development experience. Some of them you’ve already seen in other models, while others are introduced with the o1 for the very first time. These include:

Vision input: By adding features for vision input, o1 extends its reasoning ability to process and analyze visual data, extract valuable insights, and generate complex textual outputs.
Developer News: Developer messages are like system messages in GPT models, allowing instructions and context to be passed for specific use cases with the “role”: “developer” attribute.
Mental effort parameter: The reasoning effort parameter allows you to adjust the model’s cognitive load with options for low, medium, and high reasoning levels, offering more control over the model’s performance.
Extended pop-up window: The o1 model now boasts an expanded popup window of 200,000 tokens and a maximum output of 100,000 tokens, providing plenty of room for complex and detailed answers.
Structured outputs: o1 now adds support for structured output to help simplify the generation of well-defined structured output constrained by JSON schemas.
Tools: You can now use the output of the o1 model to generate input for your functions just like the previous models.
Lower latency: o1 uses on average 60% fewer logical tokens than o1-preview for a given request.

With the introduction of tooling support, AI agent applications will supercharge these models, enabling these models to plan, reason, and then implement agent-based tools to automate business processes. These newly added o1 features can offer significant benefits to developers by improving customization, control, capacity and insight into AI-based scenario reasoning. Taking advantage of these improvements in Azure OpenAI Service can ensure that developers can use the latest technologies while maintaining enterprise-grade security and global compliance.

Agility and global reach with privacy and security

At Microsoft, we are committed to bringing the latest OpenAI models to our customers with same-day access. The o1 model is no exception, offering cutting-edge reasoning capabilities now available in Azure. Our service is designed to provide a wide global reach, one of the best local data options, and can ensure your data is private and secure.

Global reach and local data: Azure OpenAI offers 99.9% reliability with access in 28 regions, ensures data residency and compliance with local regulations.
The latest innovation: We provide same-day access to the latest OpenAI models, so you’ll be ahead of the curve.
Enterprise level security: Built-in security features, including private networks, managed identity and Azure monitoring, keep your data safe.
Developer friendly integration: Seamless integration with GitHub, Visual Studio, and Copilot Studio makes it easy for developers to build and deploy AI solutions.
AI manager: Our service includes built-in responsible AI features, such as content filters, fast shields and ground detection, that support the safe and ethical use of AI.
New features with o1: The o1 model introduces developer messages, logical effort parameters, and extended response objects, improving functionality and customization for developers.

Azure OpenAI also introduces several new fine-tuning features, including support for the o1-mini model through reinforcement fine-tuning. This enables organizations to tailor AI models to their specific needs, increasing performance and reducing costs. The service also introduces direct preference optimization (DPO) to adjust model weights based on human preferences, and stored completion to capture and store input-output pairs for fine-tuning. These enhancements demonstrate a commitment to deliver robust, flexible and efficient AI solutions with Azure OpenAI.

In addition to these updates, we recently introduced a set of developer-focused features to enhance your AI development experience in your preferred environments, such as GitHub and Visual Studio Code. With Azure AI Foundry, GitHub models, the new Azure AI Agent Service, and built-in security tools like assessment and monitoring, developers can integrate advanced AI capabilities directly into their applications. These tools provide robust support for building, testing, deploying, and managing AI models, ensuring you have the flexibility and control you need to innovate and excel in AI systems you can trust. By taking advantage of these new features, you can streamline your workflows, increase efficiency, and unlock new possibilities in AI-based development.

Join us on this journey

We invite you to explore the possibilities of the o1 model and see how it can transform your business. With Azure OpenAI Service, you have access to the latest innovations, enterprise-grade security, and a global reach that ensures your data is private and secure. Join us on this journey and discover the potential of artificial intelligence to help you achieve success.

To access the latest models like o1, register with Azure AI Foundry today.