NVIDIA Jetson Orin Nano Super: Unleashing 1.7× Faster Gen AI Performance

In partnership with

A Thought Leadership platform to help the world navigate towards Artificial General Intelligence We are committed to navigate the path towards Artificial General Intelligence (AGI) by building a community of innovators, thinkers, and AI enthusiasts.

Welcome to Gen Matrix: Your Guide to GenAI Innovation and Adoption across Industries

Discover the trailblazers driving AI innovation with Gen Matrix.

Our platform showcases:

• Organizations: Industry early adopters integrating Generative AI

• Startups: GenAI innovators in hardware, infrastructure, and applications

• Leaders: Influential figures shaping the GenAI ecosystem

Why Choose Gen Matrix?

Stay ahead with our comprehensive insights into the evolving Generative AI landscape. Coming Soon: Our inaugural Gen Matrix launches December 2024. Sign up now to access the report!

Click here for the Matrix report.

TheGen.AI News

NVIDIA Jetson Orin Nano Super: Unleashing 1.7× Faster Gen AI Performance

NVIDIA has announced a significant performance upgrade for its Jetson Orin Nano Developer Kit, unveiling a "Super Mode" that boosts generative AI performance by nearly 70 percent. This upgrade is also being extended to existing Jetson Orin Nano and Orin NX modules at no additional cost.

"This is a brand-new Jetson Nano Super," said NVIDIA’s founder and CEO, Jensen Huang, in a video announcement. "It delivers almost 70 TOPS of performance at 25 watts, and it’s priced at $249. It can even run large language models. Go try it and enjoy robotics!"

While the Jetson Orin Nano Super Developer Kit retains the same hardware as the original — including the six-core Arm Cortex-A78AE CPU, Ampere-based GPU with 1,024 CUDA cores and 32 Tensor cores, and 8GB of LPDDR5 memory — NVIDIA is now unlocking previously untapped potential. By enabling "Super Mode," the GPU clock speed is nearly doubled from 625MHz to 1,050MHz, and CPU cores receive a smaller bump from 1.5GHz to 1.7GHz. This new mode increases the power target from 15W to 25W, enabling the performance boost.

The upgrade isn’t exclusive to new purchases. NVIDIA confirmed that all current Jetson Orin Nano 4GB and 8GB, as well as Jetson Orin NX 8GB and 16GB modules, will receive the Super Mode update via a software release. For Orin Nano modules, this unlocks a 25W performance mode, while Orin NX modules gain a 40W option.

In real terms, the performance improvements are substantial. The Orin Nano 4GB now delivers 32 TOPS of sparse INT8 compute and 17 TOPS dense INT8. The Orin Nano 8GB version reaches 67 TOPS sparse and 33 TOPS dense. The Orin NX range sees even greater gains: the Orin NX 8GB achieves 117 TOPS sparse and 58 TOPS dense, while the Orin NX 16GB tops out at 157 TOPS sparse and 78 TOPS dense, with boosted performance from its integrated NVIDIA Deep Learning Accelerator (NVDLA).

To sweeten the deal, NVIDIA has also slashed the price of the Jetson Orin Nano Developer Kit. Originally launched at $499, it is now available for just $249, making the upgrade more accessible for developers and robotics enthusiasts alike.

Adobe’s Vision: How Generative AI is Transforming Advertising Campaigns

Generative artificial intelligence (Gen AI), the technology behind popular tools like ChatGPT, is transforming how brands and agencies create campaigns, and this is just the beginning.

“Generative AI will be the most transformative technology of our lifetime,” said Sam Garfield, Adobe’s head of digital strategy for communications, media, and travel, in an interview at the Beet Retreat Santa Monica with Beet.TV’s Rob Williams.

Adobe, known for pioneering creative software like Photoshop and Illustrator, is now integrating Gen AI as a powerful tool for content creation across the advertising industry.

“We’re seeing tremendous opportunities in the media and entertainment sectors with generative AI,” Garfield explained. “The technology is unlocking new levels of creativity.”

One major benefit of Gen AI is its ability to produce personalized advertising at scale, which was previously too costly. Marketers can also test various creative concepts before committing significant resources.

“Generative AI empowers marketers, advertisers, and publishers to lower production barriers and dramatically increase both the volume and personalization of content,” Garfield added.

Addressing Brand Safety Concerns

The rise of Gen AI has sparked concerns among creative professionals, as seen in recent Hollywood actors and writers strikes. Adobe is tackling these challenges with a focus on transparency and brand protection.

“When we engage with media companies, they typically highlight two major concerns,” Garfield noted. “The first is rights management — ensuring the data used to train Gen AI models was ethically and legally sourced.”

The second concern is brand safety, which involves protecting advertisers from being associated with inappropriate or objectionable content.

“Adobe’s tools address this directly,” Garfield said. “Our generative AI includes safeguards to ensure compliance with brand safety guidelines, offering checks and balances to mitigate risks for media and entertainment companies.”

Google Introduces Whisk: A Visual-First Revolution in AI Image Generation

Alongside the unveiling of Veo 2, Google has launched an updated version of Imagen 3 globally and introduced Whisk, a new playful image-generation tool in Google Labs.

• Imagen 3 is now available to ImageFX users in over 100 countries, offering improved composition, richer details, and better interpretation of user prompts across diverse artistic styles, including photorealism, impressionism, and anime.

• Whisk combines Imagen 3 with Gemini's visual understanding to allow users to "remix" images by mixing and matching subjects, scenes, and styles.

• Designed for rapid ideation, Whisk helps create custom concepts like digital plush toys, enamel pins, and stickers.

The enhanced Imagen 3 model brings significant improvements in rendering brightness, textures, and user-prompt accuracy, and it is now accessible worldwide through ImageFX.

Meanwhile, Whisk takes a fresh, visual-first approach to AI image generation, moving beyond traditional text-based prompts. Instead, users can simply drag and drop reference images to define three key components: subject, scene, and style. Powered by Gemini’s visual understanding, the tool automatically generates captions from the images and uses those inputs in Imagen 3 to create new variations.

Google describes Whisk as being built for "rapid visual exploration" rather than precision editing. Early feedback from artists and designers indicates that Whisk fills a unique role in the creative process, serving as a tool for quickly exploring ideas and concepts rather than making polished edits. It is particularly useful for experimenting with designs for products like stickers, plushies, and enamel pins.

While Whisk’s results may not always match expectations—since it extracts only specific characteristics from reference images—Google encourages users to embrace the fun of the creative process. The tool also provides the underlying prompts generated by Gemini, giving users the flexibility to tweak and refine their results.

Currently, Whisk is available exclusively to users in the United States via Google Labs, adding to Google’s growing lineup of experimental AI tools.

TheOpensource.AI News

Strengthening International Partnerships to Address Open-Source AI Risks

The Center for Data Innovation convened a workshop on open-source AI risks, bringing together international experts at the Institute for Artificial Intelligence at Peking University in Beijing on December 10–11, 2024. During the event, the Center collaborated with participants to draft a statement focused on enhancing international cooperation to improve the safety and security of open-source AI systems.

“While countries compete for AI leadership, there’s no need to compete on building the foundations for AI safety,” said Daniel Castro, director of the Center for Data Innovation. “Open-source AI models and systems, being publicly accessible, offer a unique opportunity for countries to collaborate on safety initiatives. However, technical, legal, cultural, and economic barriers still limit many experts from contributing fully to global AI safety efforts. This statement outlines how governments, academia, industry, and civil society can work together to mitigate risks and maximize the benefits of open-source AI.”

The statement highlights the need to expand participation in global AI safety discussions, eliminate barriers to international collaboration, and promote inclusive engagement. It advocates for developing robust testing and evaluation methods for open-source AI, ensuring transparency in incident reporting, and prioritizing high-impact, high-likelihood risks. By pursuing these initiatives, countries can work together to address open-source AI risks while fostering its responsible and beneficial use worldwide.

How Open-Source AI is Fueling Advanced Cyber Attacks?

Open-source AI models are poised to disrupt the cyber security landscape, making advanced cyber operations accessible to a broader range of malicious actors. With the parameters of these models freely available, the risks are escalating rapidly.

AI leaders, such as Dario Amodei of Anthropic and Sam Altman of OpenAI, predict that by 2026, AI systems could surpass Nobel-level expertise across multiple domains. However, Australia’s Cyber Security Strategy, which looks ahead to 2030, touches on AI only briefly, focusing more on economic benefits than security risks. This disconnect has alarmed experts, who point to scaling laws that predict AI capabilities will advance predictably and significantly—similar to Moore’s Law for semiconductors. Massive investments, growing datasets, and more powerful infrastructure are accelerating this progress.

Recent advancements, such as OpenAI’s reasoning models like o1, show that allowing AI "thinking time" during tasks can dramatically improve performance. Giving models even more time further enhances their output, indicating rapid capability growth. While some dismiss concerns, citing issues like data bottlenecks or the belief that developers can control powerful models, these arguments fall short. Data shortages are not yet an immediate issue, and open-source availability, weak lab security, and software jailbreaks make widespread access to advanced AI nearly unavoidable.

Critics also argue that AI will benefit cyber defenders as much as attackers. However, this will only hold true if defenders recognize the scale of the problem and act swiftly. In the near future, AI systems will automate tasks currently reserved for elite cyber talent—such as probing for vulnerabilities, generating exploit code, adapting attacks, and scaling social engineering operations. Previously, sophisticated cyber weapons like Stuxnet required large, specialized teams working for months or years. With open-source AI models, a bad actor could deploy thousands of AI instances with PhD-level expertise across domains, executing complex operations at machine speed in days.

This shift could destabilize the current cyber security balance, which relies heavily on limited skilled human labor.

Despite the risks, open-source AI models also offer opportunities. They provide security researchers and Australia’s AI safety community with tools to innovate and address emerging threats. Open-source ecosystems are catching up quickly with commercial models. For instance, Meta’s Llama 3.1 405B matched GPT-4 in capability when released in July 2024. Similarly, Chinese startup DeepSeek plans to open-source its R1-Lite-Preview shortly, just months behind OpenAI’s o1-preview.

Given that the proliferation of highly capable models seems inevitable, Australia must harness these tools to its advantage.

To address these challenges, Australia must take two critical steps:

1. Engage Australia’s AI Safety Community: The national security sector should collaborate with Australia’s top AI safety experts in academia and civil society organizations, such as the Gradient Institute, Timaeus, Answer.AI, and Harmony Intelligence. By working together, they can leverage the best open-source models to analyze frontier AI capabilities, assess associated risks, and harness these tools for national security.

2. Establish an AI Safety Institute: Australia should create an AI safety institute to facilitate collaboration among government, industry, and academia. An open-source focus could position Australia uniquely, strengthening domestic AI capabilities while offering valuable contributions to its allies.

By tapping into the expertise of the AI safety community and fostering international collaboration, Australia can both mitigate risks and unlock the benefits of open-source AI for national security.

TheClosedsource.AI News

OpenAI Rolls Out o1 Reasoning Model API Access for Select Developers

OpenAI is rolling out o1, its advanced “reasoning” AI model, to its API, initially for a limited group of developers. Starting today, tier 5 developers—those spending at least $1,000 with OpenAI and with accounts older than 30 days since their first successful payment—will gain access to o1. The model replaces the previous o1-preview version available in the API.

Unlike traditional AI models, reasoning models like o1 can effectively “fact-check” themselves, reducing errors but requiring more time to generate solutions. However, this increased capability comes at a cost: OpenAI charges $15 for every ~750,000 words analyzed and $60 for every ~750,000 words generated, making it 3–4 times more expensive than GPT-4o, OpenAI’s latest non-reasoning model.

Key Features of o1 in the API:

• Function Calling: Connect the model to external data sources.

• Developer Messages: Customize tone and style instructions for the model.

• Image Analysis: Support for processing and interpreting images.

• Reasoning Effort: A new parameter that lets developers control how long the model “thinks” before responding.

This release features a post-trained version of o1, named o1-2024-12-17, which OpenAI claims improves model behavior based on user feedback compared to the earlier ChatGPT version launched two weeks ago.

“We are rolling out access incrementally while expanding usage tiers and rate limits,” OpenAI stated in a blog post.

Additional Updates:

1. Realtime API Improvements
   OpenAI also launched updated versions of its GPT-4o and GPT-4o mini models for the Realtime API, which supports low-latency, AI-driven voice responses. The new models, gpt-4o-realtime-preview-2024-12-17 and gpt-4o-mini-realtime-preview-2024-12-17, offer better data efficiency, reliability, and lower costs.

The Realtime API, still in beta, has gained new capabilities:

• Concurrent Out-of-Band Responses: Background tasks like content moderation can now run without disrupting real-time interactions.

• WebRTC Support: The API now integrates WebRTC, an open standard for real-time voice apps, enabling smoother performance on browsers, mobile devices, and IoT systems.

Notably, OpenAI hired Justin Uberti, WebRTC’s creator, in December, signaling its commitment to improving real-time AI interactions. The integration handles critical tasks like audio encoding, streaming, noise suppression, and congestion control to ensure seamless performance even on unreliable networks.

2. Preference Fine-Tuning
   OpenAI added preference fine-tuning to its fine-tuning API. This feature allows developers to compare model responses and train the AI to prioritize preferred answers over less desirable ones.

3. SDK Beta for Go and Java
   OpenAI launched official software development kits (SDKs) for Go and Java in beta, expanding tools for developers working in these programming languages.

Microsoft CEO Credits OpenAI’s Lead to Two-Year Advantage in AI Development

OpenAI is widely regarded as one of the leading AI labs globally, thanks to its early advancements and the widespread adoption of its technology, including ChatGPT. Microsoft CEO Satya Nadella recently echoed this sentiment during an episode of the BG2Pod podcast with Brad Gerstner and Bill Gurley. Nadella remarked:

“The advantage we have had, and OpenAI has had, is that we’ve had two years of runway — pretty much uncontested.”

However, Nadella also acknowledged that maintaining a significant lead in AI development over competitors may now prove challenging for OpenAI.

Microsoft and OpenAI Partnership Dynamics

Microsoft’s multi-billion-dollar partnership with OpenAI has faced scrutiny. Regulators, particularly in the UK, initially investigated whether Microsoft’s investment in OpenAI was a disguised acquisition. However, authorities ultimately ruled it was purely an investment, leading to the closure of the antitrust investigation.

Concerns about the partnership extend beyond regulators to Microsoft’s internal teams. Some insiders have expressed unease, claiming Microsoft is increasingly acting as “a glorified IT department” for OpenAI. Reports suggest that the partnership has slowed Microsoft’s AI Platform team’s progress, stalling the development of products like Azure Cognitive Search, Azure AI Bot Service, and Kinect DK. This strained relationship has even led to the departure of a senior Microsoft executive.

Despite the challenges, the partnership remains mutually beneficial. Microsoft gains access to OpenAI’s cutting-edge AI models, which are integrated across its platforms, while OpenAI benefits from Microsoft's funding and cloud computing resources to power its ambitious AI projects.

Nadella’s remarks come amid reports that leading AI labs, including OpenAI, are struggling to develop advanced AI models due to a shortage of high-quality training data. However, both former Google CEO Eric Schmidt and OpenAI CEO Sam Altman have dismissed these concerns, stating there’s no indication that scaling laws have hit a ceiling.

In a related interview, Nadella suggested that Google, despite having the resources and potential, failed to seize the leadership position in AI. This comment drew a sharp response from Google CEO Sundar Pichai, who took aim at Microsoft, saying:

“I would love to do a side-by-side comparison of Microsoft’s own models and our models any day, any time. They’re using someone else’s models.”

The rivalry highlights ongoing tensions as tech giants compete to dominate the rapidly evolving AI landscape.

Start learning AI in 2025

Everyone talks about AI, but no one has the time to learn it. So, we found the easiest way to learn AI in as little time as possible: The Rundown AI.

It's a free AI newsletter that keeps you up-to-date on the latest AI news, and teaches you how to apply it in just 5 minutes a day.

Plus, complete the quiz after signing up and they’ll recommend the best AI tools, guides, and courses – tailored to your needs.

Sign up to start learning.

Don’t miss out on the insights driving the future of Artificial Intelligence! Join a community of researchers, developers, and AI enthusiasts to stay ahead of the curve in Generative AI. Each edition delivers exclusive updates, expert analysis, and thought-provoking discussions straight to your inbox. Subscribe today and be part of the journey toward AGI innovation.

Contact us for any paid collaborations and sponsorships.

Unlock the future of problem solving with Generative AI!

If you're a professional looking to elevate your strategic insights, enhance decision-making, and redefine problem-solving with cutting-edge technologies, the Consulting in the age of Gen AI course is your gateway. Perfect for those ready to integrate Generative AI into your work and stay ahead of the curve.

In a world where AI is rapidly transforming industries, businesses need professionals and consultants who can navigate this evolving landscape. This learning experience arms you with the essential skills to leverage Generative AI for improving problem-solving, decision-making, or advising clients.

Join us and gain firsthand experience in how state-of-the-art technology can elevate your problem solving skills using GenAI to new heights. This isn’t just learning; it’s your competitive edge in an AI-driven world.