Hot Takes are occasional timely posts that focus on fast-moving news and releases, in addition to my regular Thursday and Sunday columns.
If Hot Takes aren’t your cup of tea, simply go to your account at www.whytryai.com/accountand toggle the “Notifications” settings accordingly:
TL;DR
DeepSeek released an image model called Janus-Pro-7B, which some claim makes better images than DALL-E 3 and Stable Diffusion, but it’s nowhere near in my tests.
What is it?
DeepSeek describes Janus-Pro-7B as “a novel autoregressive framework that unifies multimodal understanding and generation.”
In short, this means you can use the same model to process image inputs and generate new images. This makes Janus-Pro-7B quite flexible, combining the capabilities of task-specific models in a single one.
That—and the fact that it’s yet another open-source model—is worthy of praise.
But DeepSeek also shared a few benchmarks that show Janus-Pro-7B outperforming OpenAI’s DALL-E 3 and Stable Diffusion 3 Medium:
