DeepSeek launches Janus-Pro image generator, taking aim at DALL-E 3 superiority
Listen to article
DeepSeek, the rapidly rising AI company, has unveiled a new suite of multimodal AI models under the Janus-Pro family, which it claims can outperform OpenAI's DALL-E 3.
These models, ranging from 1 billion to 7 billion parameters in size, are available for download via the AI development platform Hugging Face.
The Janus-Pro models are licensed under the MIT license, allowing for unrestricted commercial use.
DeepSeek describes the Janus-Pro as a 'novel autoregressive framework' capable of both image analysis and creation.
According to the company, the largest Janus-Pro model, Janus-Pro-7B, outperforms DALL-E 3 and other models like PixArt-alpha, Emu3-Gen, and Stability AI's Stable Diffusion XL on AI evaluation benchmarks such as GenEval and DPG-Bench.
While some of these competing models are older, and most Janus-Pro models can only analyze smaller images (up to 384 x 384 resolution), the performance of Janus-Pro remains impressive considering its compact design.
Photo: DeepSeek
DeepSeek believes that Janus-Pro, with its simple yet powerful framework, surpasses previous unified models and challenges task-specific models. This makes it a strong contender in the field of next-generation unified multimodal models.
The company has gained widespread attention after its chatbot app rose to the top of the Apple App Store charts. Funded primarily by High-Flyer Capital Management, a quantitative trading firm, DeepSeek's language models, developed using compute-efficient methods, have raised questions about the future of AI development and whether other nations can challenge U.S. dominance in the AI sector, especially regarding AI chip demand.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Business Recorder
a day ago
- Business Recorder
OpenAI says China's Zhipu AI gaining ground amid Beijing's global AI push
OpenAI said on Wednesday its analysts have seen notable progress by Chinese start-up Zhipu AI in securing government contracts across several regions, signaling China's growing momentum in pursuing global AI leadership. Zhipu AI, which is backed by the Chinese Communist Party, aims to 'lock Chinese systems and standards into emerging markets before U.S. or European rivals can, while showcasing a 'responsible, transparent and audit-ready' Chinese AI alternative,' the ChatGPT maker said in its post. The company provides AI solutions — including sovereign large language model infrastructure and private hardware in partnership with Huawei — to governments and state-owned enterprises in Malaysia, Singapore, the UAE, Saudi Arabia and Kenya. The initiative forms part of China's broader effort to build a self-sufficient, globally competitive AI ecosystem that rivals the United States while reducing reliance on American technology. China's DeepSeek releases an update to its R1 reasoning model Zhipu AI, which did not immediately respond to a request for comment, is supported by more than $1.4 billion in state investment and maintains strong ties with the Chinese government and state-owned entities, according to OpenAI. In January, the company was added to the U.S. Commerce Department's export control entity list, barring it from procuring U.S. components. Zhipu AI has been positioning itself among China's leading AI firms alongside DeepSeek, Moonshot AI, and Minimax, as well as tech giants ByteDance and Alibaba. OpenAI has also built partnerships and attracted investment across the Middle East and Asia. Its 'OpenAI for Countries' initiative helps interested governments develop 'sovereign AI capability' in coordination with the U.S. government. The report comes as the U.S. and China engage in intense competition in the rapidly evolving AI landscape, with both nations vying for technological dominance.
Express Tribune
2 days ago
- Express Tribune
Versailles statues now talk to visitors in AI-powered experience
Curious visitors can delve into Versailles' rich history by scanning a QR code next to one of some 20 garden statues -- triggering interactive conversations in French, English, or Visitors to France's famed Palace of Versailles can now strike up a conversation with talking statues instead of listening to a traditional audio guide, as part of a new collaboration with artificial intelligence companies, the palace has said. Versailles late Monday announced a partnership with US-based OpenAI and French start-up Ask Mona to bring a modern AI touch to the iconic 17th-century palace. Curious visitors can delve into Versailles' rich history by scanning a QR code next to one of some 20 garden statues -- triggering interactive conversations in French, English, or Spanish. "The Palace of Versailles is now testing artificial intelligence, whose tremendous capabilities will greatly enrich the visitor experience," said the museum's president, Christophe Leribault. The heritage site welcomes some eight million visitors annually, with OpenAI and Ask Mona saying it was a golden opportunity to showcase their technology in a world-famous location. "Whether you're a heritage expert, a museum curator, or a visitor setting foot in the gardens of the Palace of Versailles for the first time, there's something for everyone," said Julie Lavet, OpenAI's head of European partnerships. And Ask Mona's president said this partnership is a chance to highlight AI's lesser explored applications. "Often, when we think of artificial intelligence, we think of it in terms of productivity, but here, it's really artificial intelligence that is a lever for curiosity," said Marion Carre.
Express Tribune
6 days ago
- Express Tribune
MIT AI study: Using tools like ChatGPT is making you dumber, study reveals
A new study from the Massachusetts Institute of Technology (MIT) suggests that frequent use of generative artificial intelligence (GenAI) tools, such as large language models (LLMs) like ChatGPT, may suppress cognitive engagement and memory retention. In the experiment, published by MIT, researchers monitored the brain activity of participants as they wrote essays using different resources: one group relied on LLMs, another used internet search engines, and a third worked without any digital tools. The results revealed a consistent pattern — participants who used GenAI tools displayed significantly reduced neural connectivity and recall, compared to those who relied on their own cognitive abilities. Brain scans taken during the experiment showed that LLM users exhibited weaker connections between brain regions associated with critical thinking and memory. While their essays scored well in both human and AI evaluations — often praised for their coherence and alignment with the given prompt — the writing was also described as formulaic and less original. Notably, those who used LLMs struggled to quote from or recall their own writing in subsequent sessions. Their brain activity reportedly "reset" to a novice state regarding the essay topics, a finding that strongly contrasts with participants in the "brain-only" group, who retained stronger memory and demonstrated deeper cognitive engagement throughout. Participants who used search engines showed intermediate neural activity. Though their writing lacked variety and often reflected similar phrasing, they exhibited better memory retention than the LLM group, suggesting that the process of searching and evaluating sources provided more mental stimulation. In a later phase of the experiment, the groups were shuffled. Participants who had initially used GenAI tools showed improved neural connectivity when writing without digital aids — an encouraging sign that cognitive function could rebound when AI dependence is reduced. The findings could carry important implications for education and the workplace. BREAKING: MIT just completed the first brain scan study of ChatGPT users & the results are terrifying. Turns out, AI isn't making us more productive. It's making us cognitively bankrupt. Here's what 4 months of data revealed: (hint: we've been measuring productivity all wrong) — Alex Vacca (@itsalexvacca) June 18, 2025 With GenAI tools increasingly integrated into school assignments and professional tasks, concerns about cognitive atrophy are rising. Some students now generate entire essays with tools like ChatGPT, while educators rely on similar software to grade and detect AI-generated work. The study suggests that such widespread use of digital assistance — even when indirect — may hinder mental development and reduce long-term memory retention. As schools and organisations continue to navigate the integration of AI tools, the MIT research underscores the importance of balancing convenience with cognitive engagement. Researchers suggest that while GenAI can be a useful aid, overreliance could have unintended consequences for human memory and creativity.
