• Alphawise
  • Posts
  • DEEPSEEK IMAGE GENERATION IS HERE!

DEEPSEEK IMAGE GENERATION IS HERE!

Today's featured talks include World Economic Forum debate on technology

Author

Matthew McCann
January 28, 2025

A technical AI newsletter
written with an entrepreneurial spirit for builders

What is today’s beat?

RELEASES
🧨 Open AI Dalle, meet Deepseek JANUS PRO 7B
🧨 DeepSeek faces privacy allegations, cyber attacks, and service pause


FOLLOW US ON SOCIAL FOR MORE


Your FREE newsletter
 share or subscribe 
to show support

🧨 DEEPSEEK 🧨

Open AI Dalle, meet Deepseek JANUS PRO 7B

DeepSeek AI has released Janus Pro 7B, an open-source multimodal AI model capable of processing text and images that challenges OpenAI’s DALL-E 3. The model outperforms DALL-E 3 and Stable Diffusion on several benchmarks, showcasing advancements in generative AI and democratized innovation for text-to-image systems. And yes, it’s 100% open source, again! 🎤 ⬇️ 

  • It’s Utility 
    Janus Pro 7B is a multimodal transformer with both text and vision processing capabilities.

  • Strategic Context
    The release positions DeepSeek as a competitive player in global AI innovation, particularly against U.S. companies, amid geopolitical and economic AI tensions.

  • Impact of Release: This model sparks discussions on the open-source AI ecosystem's potential to challenge proprietary models while also amplifying concerns over national security and ethical implications.

The launch of Janus Pro 7B marks a significant stride in open-source AI innovation, offering advanced multimodal capabilities at no cost. Its competitive performance against leading proprietary systems intensifies global AI rivalry while reinforcing the potential for community-driven AI development and broader accessibility.

⭐️ NEWS ⭐️

DeepSeek faces privacy allegations, cyber attacks, and service pause

Source: The Guardian

DeepSeek AI, a rising AI platform, has paused user sign-ups amid privacy concerns and a reported cyber incident. Allegations regarding data sharing with China, coupled with technical vulnerabilities, have sparked public scrutiny and industry debate over security practices. Aside from its groundbreaking released at 200x less the cost, there are other issues at stake. Let’s look at them in short-form:

  • Privacy Concerns Spark Controversy
    Allegations emerged suggesting DeepSeek may share user data with Chinese entities - this is unconfirmed but common of Chinese technologies.

  • Cyber Incident Disrupts Operations
    The platform reportedly suffered a malicious cyberattack, prompting DeepSeek to halt new user sign-ups, interrupting its release.

  • App Restrictions Tighten
    Allegedly, but we are not sure about this one, because of its non-democratic roots and heavy government influence. The heading image above says it all.

🗞️ NEWSROOM 🗞️ 

What’s hot in tech right now?

Retro Biosciences' $1B Quest to Extend Human Lifespan
Explores Retro Biosciences' mission to combat aging with cutting-edge tech, backed by $1B funding and Sam Altman's vision for longevity breakthroughs.

State of open video generation models in Diffusers
A Huggingface blog exploring the the current SOTA (state of the art) with video generation models - a great look into what was HOT in December.

Dutch startup specialising in navigation tech for mobile robotics acquired
Details SICK's acquisition of Accerion, highlighting its focus on enhancing mobile robotics with precise positioning and navigation solutions for industrial applications.

OpenAI-backed 1X acquires Kind Humanoid
Covers 1X's acquisition of Kind Humanoid, supported by OpenAI, aiming to advance humanoid robotics with AI-driven innovation and real-world applications.

⭐️ BUILDER BYTES ⭐️

What’s hot for builders right now?


Qwen2.5-1M Technical Report
The paper associated with Alibaba’s new released of its new 1M context window LLM 🤯 . View the model on modelscope or try out their chat and their technical report on Github.

Chain of Agents: LLMs Tackling Long-Context Tasks
A recent Google paper released, that goes over Google's method of using chained language models as collaborative agents to handle complex, long-context tasks efficiently with enhanced reasoning (another significant effort for reasoning).

Flows AI: Simplify AI Workflows with Vercel AI SDK
A Javascript developer must view, this introduces a library to streamline AI workflow creation atop Vercel's AI SDK, focusing on rapid integration.

LM Studio: Local LLM Deployment Made Easy
Developer must see. Highlights an open-source desktop app enabling local deployment of large language models with an intuitive interface for customization and offline AI use cases. You can run DeepSeek locally!

🤩 COMMUNITY 🤩

What’s the latest beat?

Podcast

Discusses scaling AI for scientific research at SDSC (Surgical Data Science Collective), focusing on infrastructure, computational challenges, and enabling breakthroughs in data-driven science.

TALK

World Economic Forum debate on the emerging technology. With AI, space exploration and biotechnology advancing rapidly, some see these innovations as solutions to humanity’s greatest challenges, while others raise concerns about ethics, society and inequality.

TALK

Open Source AI Infrastructure Ecosystem | AI House Davos 2025

Brain food - This stellar AI House Davos panel discussion explores the critical role of the open source AI infrastructure ecosystems. Features the following speakers:
- Bill Jia (VP Eng., Google)
- Yann LeCun (VP & Chief AI Scientist, Meta FAIR)
- Neil Lawrence (DeepMind Professor of ML, U. of Cambridge)

BLOG

Outlines practical tips for Scrum teams to enhance efficiency, collaboration, and adaptability by refining sprint planning, retrospectives, and workflows.

THANK YOU

👇️ we are 100% free, so please let us know what you think! 👇️ 

Looking to promote your company, product, service, or event?