Azure Updates: AI Foundry; GPT-image-1; Platform shift

Microsoft CVP Jessica Hawk explored emerging AI Foundry capabilities. She positioned AI Foundry “like the GPS for your AI ambitions.” Currently, there are 1,917 models available in AI Foundry’s model catalog, giving users the ability to rapidly evaluate models based on cost, throughput, quality, or industry specialization. The Evaluation API in Azure OpenAI Service allows rapid automated comparisons of OpenAI models with options to quantify improvements. Hawk gave examples of some of the models available to customers from DeepSeek R1 to Cohere Command A and Embed 4 models. The Mistral OCR model allows customers to nuance and assess hard copy documents.

On the theme of AI Foundry, Azure OpenAI Service Product Lead Steve Sweetman discussed the GPT-image-1 image generation model. “This groundbreaking model sets a new standard in generating high-quality images, solving complex prompts, and offering zero-shot capabilities in various scenarios,” he wrote. 

According to Sweetman, GPT-image-1 builds on the strengths of DALL-E with a number of enhancements like granular instruction response, text rendering, and image input acceptance. At the moment, it support a variety of features like text-to-image, image-to-image, text transformation, and inpainting. Microsoft foresees use cases around UI, game production, storybook creation, and educational material creation.