In addition to attracting market attention with its eponymous large-scale artificial intelligence model, DeepSeek earlier announced its multimodal artificial intelligence model "Janus-Pro-7B" that combines computer vision through GitHub.
Similar to multimodal models proposed by other companies, Janus-Pro-7B can also analyze image content through computer vision and provide deeper inferences and answers based on text comprehension capabilities.
According to the description, Janus-Pro-7B can analyze and describe image content, identify geographic locations, recognize text within images, and answer questions about the context of an image. Furthermore, images generated by Janus-Pro-7B can be higher quality, with more realistic details, and can even generate more appropriate images based on user input.
Like the DeepSeek R1 model, which was previously available as open source, the Janus-Pro-7B is now available as open source on GitHub.
Microsoft also announced earlier that the distilled "DeepSeek-R1-Distill-Qwen-1.5B" model can be run on devices that meet the "Copilot+ PC" design. It is expected to be available first on laptops equipped with Qualcomm Snapdragon X-series processors, and will later be available on laptops using Intel Core Ultra 200V series laptop processors and AMD Ryzen AI 9 series processors.
In addition to allowing "Copilot+ PC" design devices to use the "DeepSeek-R1-Distill-Qwen-1.5B" model, 70 billion parameter versions and 140 billion parameter versions will continue to be provided in the future. At the same time, more developers and enterprises will be allowed to use it on Microsoft's cloud service platform in Azure AI Foundry.
