OpenAI has just introduced the GPT-4o mini, a cutting-edge model that’s both highly efficient and cost-effective. Imagine a model that scores a whopping 82% on the Measuring Massive Multitask Language Understanding (MMLU) benchmark, outstripping GPT-3.5 Turbo, which managed only 70%. And that’s not all—GPT-4o mini is over 60% cheaper! With a stunning 128K context window and boosted multilingual capabilities, this model is set to revolutionize language processing worldwide.
By @pradeepviswav – Microsoft has announced the availability of GPT-4o mini, OpenAI’s fastest model, on Azure OpenAI Service. This model is available for free trial in the Azure OpenAI Studio Playground. #Microsoft #Azure #AzureOpenAI https://t.co/TSuKfFMXry
— NeowinFeed (@NeowinFeed) July 19, 2024
Now available on Azure AI, GPT-4o mini offers blazing-fast text processing. While image, audio, and video capabilities are on the horizon, you can start exploring its text prowess right away in the Azure OpenAI Studio Playground, at no cost. This model is poised to transform customer experiences, especially in streaming applications like assistants, code interpretation, and retrieval tasks. Take GitHub Copilot, for example—during testing, GPT-4o mini delivered real-time code completion suggestions, enhancing productivity with lightning speed.
But that’s not all. OpenAI has also rolled out several enhancements to the Azure OpenAI Service. These updates include bolstered safety features, expanded data residency options, and a global pay-as-you-go model, alongside significant performance upgrades. Safety is paramount, and Azure AI ensures this with default safety features for GPT-4o mini, incorporating prompt shields and protected material detection. Developers can now maximize model advancements without compromising safety, thanks to improved throughput and speed, supported by an asynchronous filter. This has been a game-changer for industries like game development, tax filing, and education, where safeguarding generative AI applications is crucial.
if you’re on Azure though, hold your horses: GPT-4o mini only available in the Playground (can’t call the API yet) https://t.co/Y5RLqLMu2W pic.twitter.com/K323Vmz0an
— Fabien Snauwaert 🇫🇷🇺🇸 🇪🇸🇭🇺🇷🇺 (@fabiensnauwaert) July 19, 2024
Azure’s data residency commitments are stronger than ever, offering customers complete control over where their data is stored and processed. Now available in 27 regions, including the recent addition of Spain, Azure AI’s data residency solutions meet diverse compliance requirements and support various hosting structures. The new global pay-as-you-go option offers flexibility for managing variable workloads, with competitive pricing at 15 cents per million input tokens and 60 cents per million output tokens. This deployment method allows seamless model upgrades within the same region, ensuring customers always have access to the latest technology.
The Azure OpenAI Service’s global pay-as-you-go deployment offers unparalleled scalability and throughput, with 15 million tokens per minute (TPM) for GPT-4o mini and 30 million TPM for GPT-4o. This service guarantees 99.99% availability, matching OpenAI’s industry-leading speed. Additionally, Azure AI’s Batch service, launching this month, will provide high-throughput job processing at significant cost savings by utilizing off-peak capacity. Fine-tuning capabilities for GPT-4o mini are also being released, allowing for customized model adjustments to meet specific use case needs with unprecedented speed and cost efficiency.
GPT-4o mini available in the early access playground for Azure OpenAI: https://t.co/3k2yc8jVJ1
— OpsConfig (@OpsConfig) July 18, 2024
More than 53,000 customers are leveraging Azure AI to create groundbreaking solutions at impressive scales. Among them are Vodafone’s customer agent solution, the University of Sydney’s AI assistants, and GigXR’s AI virtual patients. With over half of the Fortune 500 companies building applications on Azure OpenAI Service, the introduction of GPT-4o mini is set to spark further innovation and efficiency. The future is bright as customers unlock the full potential of GPT-4o mini on Azure AI.
Major Points:
- OpenAI unveils GPT-4o mini, a highly efficient model outperforming GPT-3.5 Turbo and significantly reducing costs.
- GPT-4o mini is now available on Azure AI, initially supporting text processing with future updates for image, audio, and video.
- Azure OpenAI Service enhancements include expanded safety features, data residency, and global pay-as-you-go options.
- Azure AI provides safety by default, high throughput, and scalable global deployment for GPT-4o mini.
- Over 53,000 customers, including many Fortune 500 companies, are innovating with Azure AI and GPT-4o mini.
RM Tomi – Reprinted with permission of Whatfinger News
