OpenAI today announced that the new flagship model GPT-4o mini logged in at the same time Azure AI, supports text processing capabilities and will be followed by image, audio and video capabilities.
GPT-4o mini is much smarter than GPT-3.5Turbo - scoring 82% in large-scale multitasking language understanding (MMLU) measurements versus 70% for GPT-3.5Turbo - and more than 60% cheaper.1 The model provides an extended 128K context window and integrates GPT-4o's improved multilingual capabilities. -and more than 60% cheaper.1 The model provides an extended 128K context window and integrates the improved multilingual capabilities of GPT-4o. A free trial of GPT-4o mini is available in the Azure OpenAI Studio Playground.
Microsoft Azure AI brings default security, extended data residency and service availability upgrades to GPT-4o mini. Customers can expect to experience even better performance and functionality on Azure AI, especially for streaming scenarios such as assistants, code interpreters and retrieval.
In Azure AI announced global pay-as-you-go with maximum throughput limits for GPT-4o mini. Customers now have the flexibility to pay based on resources consumed, while traffic will be routed globally to provide higher throughput, as well as static control of where data is stored. The global pay-as-you-go deployment option will provide throughput of 15 million tokens per minute (TPM), providing GPT-4o mini with 99.991 TP3T of availability and the same industry rate as OpenAI.
GPT-4o mini will be coming to Azure AI this month and will be available in the Batch service, which delivers high-throughput jobs within 24 hours at a discounted rate of 50% by using off-peak capacity. This is only possible if Microsoft is running on Azure AI, which enables Microsoft Azure AI to deliver off-peak capacity to customers.
Microsoft Azure AI is also releasing fine-tuning capabilities for GPT-4o mini this month, enabling customers to further customize models for specific use cases and scenarios. Following last month's update to token-based training billing, Microsoft Azure AI has reduced hosting costs by 43%. combined with its low inference price, this makes Azure OpenAI service fine-tuning deployments the most cost-effective offering for customers with production workloads.