February 8, 2025

GPT-4o mini, introduced by OpenAI at the moment, is offered concurrently on Azure AI, supporting textual content processing capabilities with wonderful pace and with picture, audio, and video coming later.

We’re additionally asserting security options by default for GPT-4o mini, expanded knowledge residency and repair availability, plus efficiency upgrades to Microsoft Azure OpenAI Service.

GPT-4o mini permits prospects to ship beautiful functions at a decrease price with blazing pace. GPT-4o mini is considerably smarter than GPT-3.5 Turbo—scoring 82% on Measuring Huge Multitask Language Understanding (MMLU) in comparison with 70%—and is greater than 60% cheaper.1 The mannequin delivers an expanded 128K context window and integrates the improved multilingual capabilities of GPT-4o, bringing higher high quality to languages from world wide.

GPT-4o mini, introduced by OpenAI at the moment, is offered concurrently on Azure AI, supporting textual content processing capabilities with wonderful pace and with picture, audio, and video coming later. Strive it for gratis within the Azure OpenAI Studio Playground.

We’re most excited in regards to the new buyer experiences that may be enhanced with GPT-4o mini, notably streaming situations resembling assistants, code interpreter, and retrieval which is able to profit from this mannequin’s capabilities. As an example, we noticed the unbelievable pace whereas testing GPT-4o mini on GitHub Copilot, an AI pair programmer that assists you by delivering code completion options within the tiny pauses between keystrokes, quickly updating suggestions with every new character typed.

We’re additionally asserting updates to Azure OpenAI Service, together with extending security by default for GPT-4o mini, expanded knowledge residency, and worldwide pay-as-you-go availability, plus efficiency upgrades. 

Azure AI brings security by default to GPT-4o mini

Security continues to be paramount to the productive use and belief that we and our prospects count on.

We’re happy to verify that our Azure AI Content material Security options—together with immediate shields and guarded materials detection— at the moment are ‘on by default’ so that you can use with GPT-4o mini on Azure OpenAI Service.

We now have invested in enhancing the throughput and pace of the Azure AI Content material Security capabilities—together with the introduction of an asynchronous filter—so you’ll be able to maximize the developments in mannequin pace whereas not compromising security. Azure AI Content material Security is already supporting builders throughout industries to safeguard their generative AI functions, together with recreation improvement (Unity), tax submitting (H&R Block), and schooling (South Australia Department for Education).

As well as, our Customer Copyright Commitment will apply to GPT-4o mini, giving peace of thoughts that Microsoft will defend prospects towards third-party mental property claims for output content material.

Azure AI now presents knowledge residency for all 27 areas

From day one, Azure OpenAI Service has been coated by Azure’s data residency commitments.

Azure AI gives customers both flexibility and control over where their data is stored and where their data is processed, offering a complete data residency solution that helps customers meet their unique compliance requirements. We also provide choice over the hosting structure that meets business, application, and compliance requirements. Regional pay-as-you-go and Provisioned Throughput Units (PTUs) offer control over both data processing and data storage.

We’re excited to share that Azure OpenAI Service is now available in 27 regions including Spain, which launched earlier this month as our ninth region in Europe.

Azure AI announces global pay-as-you-go with the highest throughput limits for GPT-4o mini

GPT-4o mini is now available using our global pay-as-you-go deployment at 15 cents per million enter tokens and 60 cents per million output tokens, which is considerably cheaper than earlier frontier fashions.

We’re happy to announce that the worldwide pay-as-you-go deployment possibility is mostly out there this month, permitting prospects to pay for the sources they devour, making it versatile for variable workloads, whereas visitors is routed globally to supply larger throughput, and nonetheless providing management over the place knowledge resides at relaxation.

Moreover, we acknowledge that one of many challenges prospects face with new fashions just isn’t having the ability to improve between mannequin variations in the identical area as their current deployments. Now, with world pay-as-you-go deployments, prospects will have the ability to improve from current fashions to the newest fashions.

World pay-as-you-go presents prospects the best attainable scale, providing 15M tokens per minute (TPM) throughput for GPT-4o mini and 30M TPM throughput for GPT-4o. Azure OpenAI Service presents GPT-4o mini with 99.99% availability and the identical business main pace as our companion OpenAI.

Azure AI presents main efficiency and suppleness for GPT-4o mini

Azure AI is constant to put money into driving efficiencies for AI workloads throughout Azure OpenAI Service.

GPT-4o mini involves Azure AI with availability on our Batch service this month. Batch delivers excessive throughput jobs with a 24-hour turnaround at a 50% low cost price by utilizing off-peak capability. That is solely attainable as a result of Microsoft runs on Azure AI, which permits us to make off-peak capability out there to prospects.

We’re additionally releasing fine-tuning for GPT-4o mini this month which permits prospects to additional customise the mannequin on your particular use case and state of affairs to ship distinctive worth and high quality at unprecedented speeds. Following our replace final month to change to token based billing for training, we’ve reduced the hosting charges by as much as 43%. Paired with our low value for inferencing, this makes Azure OpenAI Service fine-tuned deployments probably the most cost-effective providing for purchasers with manufacturing workloads.

With greater than 53,000 prospects turning to Azure AI to ship breakthrough experiences at spectacular scale, we’re excited to see the innovation from firms like Vodafone (buyer agent answer), the University of Sydney (AI assistants), and GigXR (AI digital sufferers). Greater than 50% of the Fortune 500 are constructing their functions with Azure OpenAI Service.

We will’t wait to see what our prospects do with GPT-4o mini on Azure AI!


1GPT-4o mini: advancing cost-efficient intelligence | OpenAI