How AI Models Are Becoming More Accessible

BREAKING: OpenAI’s latest models introduce a new standard for open-source reasoning systems. They have released two Mixture of Experts models under the Apache 2.0 license: gpt-oss-20b and gpt-oss-120b. Both are built specifically for tool use, advanced reasoning, and integration into agent-based workflows. Key insights: 1. Open access with strong performance: These models are fully open-weight and match or exceed the performance of commercial options such as o3-mini and 04-mini. The 120B model surpasses o3-mini on standard benchmarks including MMLU, GPQA, and code generation tasks. 2. Efficient deployment across hardware: The 20B model is small enough to run on edge devices and consumer-grade hardware. Both models support over 130,000 tokens of context and use Mixture of Experts routing to reduce compute costs during inference. 3. Advanced tool interaction capabilities: Both models are capable of fetching current information from the web, executing Python code within a notebook-style environment, and calling custom functions defined by the user. 4. Customizable reasoning depth: Users can adjust the level of reasoning between low, medium, and high depending on the complexity of the task and the desired response speed. This allows for dynamic control in agentic applications. 5. Seamless integration with deployment platforms: OpenAI has collaborated with several infrastructure providers to ensure these models work immediately across a wide range of systems, making them accessible to developers without the need for extensive setup. 6. Structured interaction format: The models use a harmony chat format that supports interleaving reasoning with tool execution. This enhances performance in multi-step, tool-augmented tasks. Have you used it yet?

A 9 billion parameter AI model is now beating a 120 billion parameter LLM on major benchmarks. It runs on a laptop. And it's open source. That's the AI story you should be paying attention to today. Alibaba's Qwen team just dropped four new models, from 0.8B to 9B parameters. The 9B version is outscoring much larger 120B models on many angles. An AI model that's 13x smaller, beating it's much more resource-hungry LLM cousin. Free. Open source. Apache 2.0 license. You can download it right now. And these aren't just text models. They're natively multimodal. The 9B runs on a single consumer-grade GPU at full precision. Grab the 4-bit model and you can run it on a 3-year old macbook. The 0.8B model runs nicely on an iphone or raspberry pi. Here's why this matters: For the last two years, the AI conversation has been dominated by who can build the biggest model, who can raise the most money for GPU clusters, and who can win the next capability benchmark. That race isn't over. But a different race has started. The race to figure out how small you can go while keeping the performance that actually matters for production use cases. If a 9B model can match a 120B model on reasoning and vision tasks, the economics of deploying AI just changed. That opens up a different business model that a whole new set of companies can play. The frontier models will keep getting bigger, better, and more capable. I'm excited to see where leaders like like Anthropic OpenAI Google and others are going with their models. But the most interesting story in AI right now might be how small the useful models are getting. Watch this space.

Microsoft’s “1‑Bit” AI Breakthrough Could Shrink AI to Your Desktop A Leaner, Simpler Future for AI Microsoft’s General Artificial Intelligence group has unveiled a groundbreaking AI model called BitNet b1.58, a neural network that uses only three weight values: -1, 0, and 1. This “1-bit” or ternary approach radically reduces the memory and processing requirements of traditional large language models (LLMs). In a field known for its massive GPU-powered clusters, BitNet’s ability to run on an everyday CPU could mark a turning point in the accessibility and sustainability of advanced AI systems. How BitNet Reinvents AI Efficiency • Ternary Weights: • Unlike conventional LLMs that use 16- or 32-bit floating point numbers for weights, BitNet uses just three values. • This simplifies calculations, enabling lightweight inference and training with significantly reduced memory usage. • CPU-Friendly Performance: • The model runs entirely on standard CPUs—no specialized GPU or cloud infrastructure required. • Opens the door for powerful AI applications on local machines and edge devices. • Model Size and Power: • BitNet b1.58 2B4T (2 billion parameters, 4 transformer blocks) is small by modern LLM standards. • Still, it performs competitively with much larger open-weight models, proving that high efficiency does not mean sacrificing capability. • Foundation in Prior Research: • Builds on Microsoft’s 2023 work on quantized models and neural scaling laws. • Shows that reducing model complexity doesn’t necessarily limit performance—especially in well-structured transformer models. Why This Matters • Democratizing AI Access: • BitNet could enable developers and researchers to experiment with powerful AI without expensive hardware. • Especially valuable for low-resource settings, classrooms, and personal computing environments. • Environmental and Cost Impact: • Traditional LLMs require energy-hungry data centers. • CPU-based models drastically cut the carbon footprint and financial barriers of deploying AI. • Edge and Offline Use: • Ideal for situations where internet access or cloud compute is limited or unavailable. • Could drive AI adoption in healthcare, agriculture, and remote field operations. Microsoft’s BitNet shows that the future of AI doesn’t have to be bigger—it can be smarter. As the tech industry grapples with cost, energy, and scalability concerns, this “1-bit” model signals a leaner, more inclusive path forward in AI innovation.

The unveiling of Deepseek-R1 has sent shockwaves through the AI industry, challenging long-held assumptions about the cost and accessibility of advanced AI models. This serves as a powerful reminder of a fundamental business principle: costs and margins matter, even in cutting-edge technology. Deepseek-R1's ability to rival or surpass leading AI models at a fraction of the cost is not just a technological feat; it demonstrates that AI initiatives must generate real value and sustainable margins, not just push technological boundaries for their own sake. This bodes well to accelerate enterprise adoption of AI and AI-driven performance improvement projects. By dramatically lowering the barrier to entry, Deepseek-R1 opens the door for businesses of all sizes to leverage advanced AI capabilities. We've seen this pattern before - when technology becomes more affordable and accessible, innovation and adoption surge. Think of the impact of the personal computer or the rise of cloud computing. Deepseek-R1 could be the catalyst for a similar transformation in AI. By offering comparable performance to leading models at a fraction of the cost, it challenges the notion that only tech giants can play in the advanced AI space. This could lead to a new wave of AI-powered solutions across industries, from healthcare and finance to manufacturing and retail. For enterprises, this means the opportunity to implement AI-driven performance improvements without breaking the bank. It's not just about cost savings, though. The real value lies in the potential for widespread experimentation and innovation. When the stakes are lower, companies are more likely to take risks and explore new applications of AI technology. As we enter this new era of accessible AI, the focus will shift from who has the biggest AI budget to who can most effectively leverage these tools to drive business value. It's a reminder that in technology, as in business, sustainable success is built on a foundation of sound economics and practical application. The Deepseek-R1 story is still unfolding, but one thing is clear: it's time for enterprises to reevaluate their AI strategies. The question is no longer "Can we afford to implement AI?" but rather "Can we afford not to?"

The 2025 Landscape of LLMs — Updated View of the Big Players in the Game of AI About 18 months ago, I shared my first version of the Large Language Model landscape, and a lot has changed since then. The space has evolved rapidly, but at the same time, we’re starting to see clear patterns emerge. This updated view focuses on the leading AI research labs, their latest models, and how those models can be accessed. It’s not meant to list every single LLM out there—but it does cover about 95% of what’s being used in real-world scenarios today. Here are some key insights: 🔹 No more clear front-runner: We’ve gone from “everyone chasing one leader” to a fairly even playing field. For most use cases, model differences are small and often not that relevant. 🔹 Model choice is the new normal: Customers now expect the ability to test, compare, and switch between models with ease. This shift is driving interest in evaluation frameworks and model routing tools. 🔹 Reasoning-first models are rising: Many providers are clearly moving toward models optimized for reasoning—fueling the surge of Agentic AI architectures. 🔹 Proprietary still leads, but just barely: Open-source and open-weight models are quickly closing the gap. 🔹 The U.S. is still ahead, but international competition is heating up—fast. 🔹 Cloud and APIs dominate: With few exceptions (hello Grok/XAI 👀), nearly every model is accessible via API across the major cloud platforms. 🔹 Serverless is the default: Most organizations prefer calling models via API over hosting or fine-tuning them—unless the use case is highly specialized. 🔹 Everyone else? Still less than 5% of the market. We’re entering a phase where model access, interoperability, and orchestration matter more than the model itself. And this landscape helps make sense of where we are and where we’re going. #LLMs #AI #MachineLearning #GenerativeAI #AgenticAI #OpenSource

You’re going to be hearing a lot about Open-source v Open-weights. Open-source AI models and "Open-weights" are distinct approaches to transparency and accessibility in AI. So what’s the difference? Open-source AI models refer to machine learning models whose entire codebase, including architecture, training pipeline, and often training data, is publicly available. This approach offers full transparency of the model's structure and training process, allows for reproduction and improvement of the model, often includes the training data, and typically comes with an open-source licence. Examples include BERT by Google and GPT-2 by OpenAI. "Open-weights", on the other hand, is the practice of releasing only the trained parameters (primarily weights) of an AI model, without necessarily disclosing the full architecture, training code, or data. This approach provides access to the model's learned parameters, allows for fine-tuning and adaptation of the model, does not necessarily reveal the full model architecture or training process, and may come with specific usage restrictions. Open-source and Open-weights differ in several key aspects. Open-source models offer comprehensive transparency, revealing the entire process from architecture to training, while openweights provide partial transparency, focusing on the end result rather than the process. Open-source models can be fully reproduced, whereas open-weights models can only be used or fine-tuned, as the training process isn't disclosed. Open-source models typically grant more extensive rights to users, while open-weights may come with more restrictions on use and modification. Open-weights can be easier to implement quickly, as users don't need to understand or reproduce the training process. Open-source models allow for more fundamental innovations in architecture and training, while open-weights focus innovation on applications and fine-tuning of existing architectures. Reproducing open-source models often requires significant computational resources for training, whereas using openweights is generally less resource-intensive, as the costly training process is already complete. From a commercial perspective, open-source models may pose challenges for companies wanting to maintain a competitive edge, while open-weights can strike a balance between openness and protecting proprietary aspects of the training process. The choice between these approaches often depends on balancing innovation fostering, intellectual property protection, and democratising access to advanced AI capabilities. By sharing learned parameters, open weights enable users to leverage and build upon sophisticated AI models without extensive computational resources or access to large training datasets. This approach has gained traction, especially with large language models, as it allows for wider use and adaptation of powerful AI systems while maintaining some level of proprietary advantage for the original developers.

The AI landscape is undergoing a fundamental shift, and it’s not the one you think. The competitive frontier isn’t only about building the largest proprietary models. There are two other major trends emerging that haven’t had enough discussion: Open source models have made tremendous strides, especially on cost relative to performance. Compact, fit-for-purpose models can meaningfully out-perform general purpose LLMs on specific tasks and do so at dramatically lower cost. Our Chief Technology Officer and AI team share how we are using open source AI models at Pinterest to achieve similar performance at less than 10% of the cost of leading, proprietary AI models. They also share how Pinterest has built in-house, fit-for-purpose models that are able to significantly outperform leading, proprietary general purpose models. The race to build the largest, most powerful models is profound and meaningful. If you want to see a thriving ecosystem of innovation in an AI-driven world, you should also want to see a thriving open source AI community that creates democratization and transparency. It’s a good thing for us all that open source is in the race. For our part, we’ll continue to share our findings in leveraging open source AI so that more companies and builders can benefit from the democratizing effect of open source AI. https://lnkd.in/gmT6UNXs

🚀 Over the past year, I’ve been sharing a simple but crucial point: the future of AI isn’t just about building the most powerful models; it’s about how effectively we use AI in real-world problems. Businesses are starting to care more about ROI, ease of implementation, and customer experience than just raw performance. Think of it like buying a car. While we might all dream of a Ferrari or Bugatti, the final decision often depends on practical factors—budget, maintenance costs, and how well the car fits the terrain and roads we use. In the same way, businesses are now asking: "Do we really need the most advanced AI model, or can we achieve our goals with something simpler and more cost-effective?" 📍 A recent example of this shift is #DeepSeek R1, a Chinese AI model that delivers excellent performance at a fraction of the cost and development time of its larger competitors. Built using fewer and less advanced chips, DeepSeek R1 is already competing with models from giants like #OpenAI and #Google. In fact, it has become the most downloaded model on platforms like #HuggingFace, showing how innovation can thrive even with limited resources. This is a reminder of how scarcity drives creativity. Startups like DeepSeek are showing the world that you don’t always need expensive, cutting-edge tools to achieve great results. They’ve rethought the process, prioritized efficiency, and built models that balance performance with cost. ❗ For businesses, this is exciting: It means they can experiment more, implement AI faster, and still get a positive return on investment. But it also challenges us to think carefully about the choices we make in AI adoption. 👉 Should we always chase the most powerful tools, or should we optimize for what we need? 👉 How do we balance innovation and affordability? 👉 And how do developments like these impact the broader AI ecosystem, including the demand for advanced chips from companies like Nvidia? I strongly believe that we are entering a phase where AI isn’t just about pushing boundaries but making those boundaries accessible to everyone. I write about #artificialintelligence | #technology | #startups | #mentoring | #leadership | #financialindependence PS: All views are personal Vignesh Kumar

Unlocking the Potential of AI in Emerging Economies: Reflections on the UN’s High-Level Report Over the last 5 days at UNGA in NY, I've had the privilege of giving 4 keynotes on the intersections of AI and AMR, Misinformation, Climate and even pandemic reliance. I've been wowed by the frontier tech on display - by Google, Meta, OpenAI, and many others! AI and DPI (Digital Public Infrastructure) are in nearly every discussion, amplified by the new Global Digital Compact... The UN’s report "Governing AI for Humanity" from the High-level Advisory Body on AI is a milestone. While AI has long been used in global health for tasks like diagnostic imaging and predicting sepsis, fast-evolving large language and multimodal models —require us to rethink how we govern, assess, and implement AI. WHO has emphasized the importance of ethical and regulatory frameworks to ensure AI is used responsibly. Our guidance (below) focuses on creating AI that is technically robust, culturally relevant and contextually appropriate for diverse settings. AI must be designed to work in the unique environments where it will be deployed - ADDRESSING priorities on the ground... How do we make AI tools more accessible? Open-sourcing models is a good start, but it’s not enough. Models should be validated in real-world settings, under the conditions typical of many LMICs—where low bandwidth, intermittent connectivity, and limited access to advanced compute are common obstacles. AI systems have to operate effectively within these constraints and we have to develop the necessary infrastructure to enable continuous evaluation and fine-tuning. We have to move beyond the notion that LMICs need to gather more data before they can fully engage with AI. The truth is, we don’t need a perfect starting point—and we will likely never have one! Foundational AI models are designed to learn and evolve. It’s up to us to create systems that allow these models to be refined and adapted to local contexts, with appropriate safeguards. Waiting only risks widening the digital divide and leaving many countries behind in the global race for AI innovation. We have to shift focus from simply validating AI models to validating the entire process of using AI in health. Systems are dynamic and evolving, and we need to be just as agile in how we monitor their deployment. We're not there yet with the tools and benchmarking that's needed, but we're working on it! So, what’s next? We must act quickly. Invest in the necessary infrastructure, such as computing power—not just for training models, but for deploying them where they are needed. Support large-scale collaborations that build systems in a sustainable and inclusive way. Foster strong partnerships across governments, academia, and the private sector to ensure transparency and accountability. #AIforHealth #DigitalHealth #LMICs #UNGA79 Nick Martin Bilal A Mateen Annie Hartley Sameer Pujari Rubayat Khan Rebecca Distler Fred Hersch Trevor Mundel

The AI arms race is missing the real opportunity: While everyone talks about billion-parameter AI models, the real revolution might be happening with smaller, specialized models. Here's why: 𝟭. 𝗠𝗼𝗿𝗲 𝗮𝗰𝗰𝗲𝘀𝘀𝗶𝗯𝗹𝗲 Smaller models can be run by more companies on more attainable hardware 𝟮. 𝗠𝗼𝗿𝗲 𝗰𝗼𝘀𝘁-𝗲𝗳𝗳𝗲𝗰𝘁𝗶𝘃𝗲 Training and running smaller models costs a fraction of massive models 𝟯. 𝗠𝗼𝗿𝗲 𝗲𝘅𝗽𝗹𝗮𝗶𝗻𝗮𝗯𝗹𝗲 The larger the model, the harder it is to understand how decisions are made 𝟰. 𝗠𝗼𝗿𝗲 𝘁𝗮𝗿𝗴𝗲𝘁𝗲𝗱 Domain-specific models excel at industry-specific language and contexts As GPU shortages continue and cloud costs increase, the companies gaining competitive advantage aren't chasing the biggest models. They're building smarter, targeted AI systems that deliver specific business value from their unstructured data. The AI revolution isn't just about size. It's about relevance.