DeepSeek launches Janus-Pro-7B model, outperforms OpenAI’s DALL-E 3 and Stable Diffusion

While the world is buzzing about DeepSeek’s R1 reasoning model, the Chinese AI startup has quietly unveiled another game-changer: Janus-Pro-7B, a multimodal AI model that generates images and outperforms OpenAI’s DALL-E 3 and Stable Diffusion on benchmarks like GenEval and DPG-Bench.
Built on DeepSeek’s LLM architecture, Janus-Pro introduces an innovative autoregressive framework that decouples visual encoding for understanding and generation, resolving conflicts in previous models. This approach enhances flexibility and performance, making Janus-Pro a top contender for next-gen multimodal AI.
Licensed under MIT, Janus-Pro is not only powerful but also accessible, continuing DeepSeek’s trend of delivering high-performing, cost-effective AI solutions. This comes just weeks after the launch of DeepSeek-R1, which is already challenging OpenAI’s dominance in reasoning models.
Is DeepSeek redefining the future of AI? Watch as we explore how Janus-Pro is pushing the boundaries of multimodal AI and what it means for the industry.
Read more at: techstartups.c...

[@AI-Heaven-Hell]

The subject is interesting. But the contentv is too clearly ai generated.