Introducing GPT-4 in Azure OpenAI Service

At Microsoft, we are constantly discovering new ways to unleash creativity, unlock productivity, and uplevel skills so that more people can benefit from using AI. This is allowing our customers to build the future faster and more responsibly by powering their apps using large-scale AI models. Our collaboration with OpenAI, along with the power of Azure have been core to our journey.

Today, we are excited to announce that GPT-4 is available in preview in Azure OpenAI Service. Customers and partners already using Azure OpenAI Service can apply for access to GPT-4 and start building with OpenAI’s most advanced model yet. With this milestone, we are proud to bring the world’s most advanced AI models—including GPT-3.5, ChatGPT, and DALL•E 2—to Azure customers, backed by Azure AI-optimized infrastructure, enterprise-readiness, compliance, data security, and privacy controls, along with many integrations with other Azure services.

Customers can begin applying for access to GPT-4 today. Billing for all GPT-4 usage begins April 1, 2023, at the following prices:

GPT-4

Prompt

Completion

8k context

$0.03 per 1,000 tokens

$0.06 per 1,000 tokens

32k context

$0.06 per 1,000 tokens

$0.12 per 1,000 tokens

GPT-4 for every business

While the recently announced new Bing and Microsoft 365 Copilot products are already powered by GPT-4, today’s announcement allows businesses to take advantage of the same underlying advanced models to build their own applications leveraging Azure OpenAI Service.

With generative AI technologies, we are unlocking new efficiencies for businesses in every industry. For instance, see how Azure OpenAI Service can allow bot developers to create virtual assistants in minutes using natural language with Copilot in Power Virtual Agents.

GPT-4 has the potential to take this experience to a whole new level using its broader knowledge, problem-solving abilities, and domain expertise. With GPT-4 in Azure OpenAI Service, businesses can streamline communications internally as well as with their customers, using a model with additional safety investments to reduce harmful outputs.

Companies of all sizes are putting Azure AI to work for them, many deploying language models into production using Azure OpenAI Service, and knowing that the service is backed by the unique supercomputing and enterprise capabilities of Azure. Solutions include improving customer experiences end-to-end, summarizing long-form content, helping write software, and even reducing risk by predicting the right tax data.

Customers are accelerating the adoption of language models

We are just scratching the surface with generative AI technologies and are working to enable our customers to responsibly adopt Azure OpenAI Service to bring real impact. With GPT-4, Epic Healthcare, Coursera, and Coca-Cola plan to use this advancement in unique ways:

"Our investigation of GPT-4 has shown tremendous potential for its use in healthcare. We'll use it to help physicians and nurses spend less time at the keyboard and to help them investigate data in more conversational, easy-to-use ways."—Seth Hain, Senior Vice President of Research and Development at Epic

"Coursera is using Azure OpenAI Service to create a new AI-powered learning experience on its platform, enabling learners to get high-quality and personalized support throughout their learning journeys. Together, Azure OpenAI Service and the new GPT-4 model will help millions around the world learn even more effectively on Coursera."—Mustafa Furniturewala, Senior Vice President of Engineering at Coursera

"Words cannot express the excitement and gratitude we feel as a consumer package goods company for the boundless opportunities that Azure OpenAI has presented us. With Azure Cognitive Services at the heart of our digital services framework, we have harnessed the transformative power of OpenAI's text and image generation models to solve business problems and build a knowledge hub. But it is the sheer potential of OpenAI's upcoming GPT-4 multimodal capabilities that truly fills us with awe and wonder. The possibilities for marketing, advertising, public relations, and customer relations are endless, and we cannot wait to be at the forefront of this revolutionary technology. We know that our success is not just about technology but also about having the right enterprise features in place. That's why we're proud to have a long-standing partnership with Microsoft Azure, ensuring that we have all the tools we need to deliver exceptional experiences to our customers. Azure OpenAI is more than just cutting-edge technology—it's a true game-changer, and we're honored to be a part of this incredible journey."—Lokesh Reddy Vangala, Senior Director of Engineering, Data and AI, The Coca-Cola Company

Our commitment to responsible AI

As we described in my previous blog, Microsoft has a layered approach for generative models, guided by Microsoft’s Responsible AI Principles. In Azure OpenAI, an integrated safety system provides protection from undesirable inputs and outputs and monitors for misuse. On top of that, we provide guidance and best practices for customers to responsibly build applications using these models, and we expect customers to comply with the Azure OpenAI Code of Conduct. With GPT-4, new research advances from OpenAI have enabled an additional layer of protection. Guided by human feedback, safety is built directly into the GPT-4 model, which enables the model to be more effective at handling harmful inputs, thereby reducing the likelihood that the model will generate a harmful response.

Getting started with GPT-4 in Azure OpenAI Service

Apply for access to GPT-4 by completing this form.
Learn more about Azure OpenAI Service and more about all the latest enhancements.
Get started with GPT-4 in Azure OpenAI Service in Microsoft Learn.
Read our Partner announcement blog, Empowering partners to develop AI-powered apps and experiences with ChatGPT in Azure OpenAI Service.
Learn how to use the new Chat Completions API (preview) and model versions for ChatGPT and GPT-4 models in Azure OpenAI Service.

Quelle: Azure