In December 2023, Microsoft announced that it had launched a public preview of OpenAI’s GPT-4 Turbo with Vision large language model in its Azure OpenAI Service. Today, the company announced that GPT-4 Turbo with Vision is now generally available for its Azure OpenAI Service customers,
In a blog post, Microsoft stated that customers in its Sweden Central and East US 2 Azure OpenAI regions can deploy the new “gpt-4-turbo-2024-04-09” model. The blog adds:
Our customers and partners have been utilizing GPT-4 Turbo with Vision to create new processes, enhance efficiencies, and innovate within their businesses. Applications range from retailers improving the online shopping experience, to media and entertainment companies enriching digital asset management, and various organizations deriving insights from charts and diagrams
Some features, such as Optical Character Recognition (OCR), object grounding, video prompts, and “Azure OpenAI Service on your data with images”, that were available in the public preview model, are currently not available on the generally available version. Microsoft stated that it is “actively exploring how to best integrate these features” for future updates of the model.
Microsoft also added that “JSON mode and function calling for inference requests involving image (vision) inputs” will be added to the new model for Azure OpenAI Service customers in the near future.
The company previously announced that pricing for GPT-4 Turbo with Vision for Azure OpenAI Service customers will be $0.01 per 1000 tokens for Input and $0.03 per 1000 tokens for Output, with different pricing for enhanced features.
Microsoft recently announced it was testing a new “Risks & safety monitoring” feature in Azure OpenAI Service. It’s designed to give businesses a real-time look at how their AI systems are operating. It also gives them a way to see if people, such as hackers, are trying to abuse those same systems.