Use a vision-capable model in the Microsoft Foundry portal
To handle prompts that include images, you need to deploy a multimodal generative AI model - in other words, a model that supports not only text-based input, but image-based (and in some cases, audio-based) input as well. Multimodal models available in Microsoft Foundry include (among others):
- Microsoft Phi-4-multimodal-instruct
- OpenAI gpt-4.1
- OpenAI gpt-4.1-mini
Tip
To learn more about available models in Microsoft Foundry, see the Microsoft Foundry Models overview article in the Microsoft Foundry documentation.
Testing multimodal models with image-based prompts
After deploying a multimodal model, you can test it in the chat playground in Microsoft Foundry portal.
In the chat playground, you can upload an image from a local file and add text to the message to elicit a response from a multimodal model.