Hook

Deepseek just dropped another AI model that can now generate images. So the new model is called Janus-Pro-7B and it's multimodal, meaning that it can now generate images. Plus, it beats out Open AI on all benchmarks once again. Looking at the GenEval and DPG benchmarks, we can see that Janus-Pro-7B, which is Deepseek's new model, is leading in all text-to-image generation models. And once again, we can see the average performance beats out every single text-to-image model. And just like all their other models, they've made it completely open source again. Basically, Deepseek has the most powerful image AI model in the entire industry. And this news gets even more wild because just this morning, Deepseek essentially caused an AI market crash within the United States because of how cheap and powerful their newest R1 model was. But they did it again today, targeting AI image generation. The AI race is now at full force between the US and China, and it really seems like Deepseek is not letting up. And at the end of the day, this is a win for us because the model is once again open source. Plus, all this competition is going to force Open AI to step on the pedal and innovate even more. And make sure to follow for more AI news and content.
Their other posts in the index, biggest breakout first.