Meta’s launch of two artificial intelligence (AI) Llama 4 models has positioned the U.S. as a leader in the AI race, David Sacks, the U.S. AI and crypto czar, said in an X post on Saturday. He wrote:

“For the U.S. to win the AI race, we have to win in open source too, and Llama 4 puts us back in the lead.”

On April 6, Meta announced the launch of its fourth-generation open-source Llama 4 models — Llama 4 Scout and Llama 4 Maverick.

The AI race intensified with the launch of Deepseek

DeepSeek, a Chinese AI startup founded in 2023, launched its first model in December 2024. In January 2025, it launched a chatbot that claimed to rival the capabilities of OpenAI’s ChatGPT.

Downloads of DeepSeek’s generative AI, DeepSeek R1, topped the charts of app stores. It shook the world’s primary assumption: Hefty investments and expensive chips are the only way to get ahead in the AI game.

As a result, the stock prices of U.S. tech companies like Nvidia suffered losses following DeepSeek R1’s launch.

This is because Deepseek claimed to have spent approximately $6 million to train its AI model. On the other hand, OpenAI reportedly spent around $100 million to train ChatGPT-4.

While venture capitalist Marc Andreessen called DeepSeek’s R1 launch “AI’s Sputnik moment,” U.S. President Donald J Trump called it a “wake-up call” for American firms.

Since then, U.S. firms have been trying to leapfrog in the AI race, and according to Sacks, who has been a vocal proponent of AI, Llama 4 has been the key.

Meta claims Llama 4 models are “best in their class”

Meta claims that the Llama 4 models are their “most advanced models yet” and also “best in their class for multimodality.” Both models are currently available for download and use on Meta applications like WhatsApp and Instagram.

Multimodal AI systems are capable of processing various types of data — text, image, audio, and video — simultaneously. This enables the AI to comprehend complex scenarios and generate comprehensive responses.

Llama 4 Scout and Llama 4 Maverick are the first open-source Meta AI models built using a mixture of experts (MoE) architecture. In an MoE, multiple smaller models or specialized experts collaborate to make the larger AI model work. This means that experts focus on solving the parts of the problem they are designed to handle.

Llama 4 Scout has 17 billion active parameters and 16 experts. Llama 4 Maverick has the same number of parameters but is designed with 128 experts. While the former can fit in a single NVIDIA H100 GPU, the latter requires an H100 host.

Meta claims that Llama 4 Scout outperforms Gemma 3, Gemini 2.0 Flash-Lite, and Mistral 3.1 across a broad range of widely reported benchmarks.

Llama 4 Maverick, on the other hand, provides results comparable to DeepSeek v3 on reasoning and coding despite having less than half the active parameters. Meta also asserts that Llama 4 Maverick beats GPT-4o and Gemini 2.0 Flash across a number of benchmarks.

Furthermore, according to Meta’s testing, Llama 4 “responds with strong political lean at a rate comparable to Grok.”

Meta also unveiled Llama 4 Behemoth, which is still in training, as one of the “world’s smartest” large language models (LLMs).

Meta launched its first Llama model in February 2023.

Mentioned in this article

Only a few weeks after preorders opened for the rebooted Pebble Core watches, Scottish indie brand UNA kickstarted “the world’s first” repairable & upgradable GPS sports watch: the UNA Watch. And I’m basking in this sudden new trend of open-source watches as the big-name brands trend toward subscriptions.

The UNA Watch Kickstarter lists out its features: continuous heart rate, blood oxygen, steps, and floors climbed in normal settings, plus fitness features like HR zones, pace, laps, and (after launch) downloadable GPX routes. It has dual-frequency GPS and 10-day battery life.

Interestingly, it runs on the same Cortex-M33 coprocessor found in the Pixel Watch 3, only without the battery-guzzling Snapdragon.

I appreciate that the UNA Watch charges with a simple USB-C slot instead of a proprietary cable and that you’ll be able to buy modular upgrades like music storage or a mic/speaker combo and install them yourself. It also won’t have any subscription.

You can even buy the dev tool for an extra $40 to add your own apps and firmware!

Pebble is much less fitness-focused, making UNA a better fit for my lifestyle. However, both share an open-source, enthusiast philosophy where people can code their own apps or add their own sensors as needed. In theory, the community will build up the UNA Watch’s limited feature set — just as Pebble asked its community for help developing new software.

My limited coding background makes it unlikely that I can help either community. But smartwatch brands have moved away from letting indie devs into their closed systems, to everyone’s detriment. That’s why devices like UNA and Pebble are so important and have me excited to buy into their communities.

UNA and Pebble are fighting against the tide

I’ve reviewed almost every mainstream smartwatch brand out there today, and it feels like they’re all starting to converge on the same priorities.

Some wearable brands are honing in on offering medical-grade health tracking and adding new data points like blood pressure and glucose. Others, like Garmin, are going all-in on AI as a pretense to charge more for premium features with its Connect Plus subscription.

Basically, we’re entering the heavyweight era of smartwatches, where brands bring in doctors for more accurate data and squeeze LLMs onto your wrist. And while that’s exciting in its way, it doesn’t leave much room for experimentation — for smaller brands, tiny teams, or indie devs to try something new — because every watch needs a laundry list of features to compete.

Part of that is Apple’s fault. I’ve previously written about how Apple’s locked watch ecosystem forced wearable brands to focus on fitness and how brands like Pebble today can’t offer messaging, third-party app stores, or other smarts because of Apple’s restrictions. Until the EU forces Apple to open up its ecosystem, that won’t change (on iOS, at least).

That’s why I’m considering dropping money on the UNA Watch or Pebble Watch, even though they’re objectively less feature-packed than other watches on the market. To support brands that will go off the beaten industry path and try new things.

More modular smartwatches, please

I’ve seen other “hackable” watches recently like Spectra that count on their community to come up with exciting ideas. But the UNA Watch is especially intriguing to me, assuming it lives up to the brand’s promises.

I appreciate that you can take out the battery and replace it yourself. A lot of wearable tech isn’t repairable — especially smart rings — and when your watch naturally loses capacity, manufacturers would rather that you buy a new watch than keep your old device chugging along.

With the UNA Watch, you can unscrew everything yourself, either to replace things or add new features years later. Based on the Kickstarter comments, they’ll choose what new modules to make based on what the community asks for, such as the mic/speaker for Bluetooth calling.

Pebble originally tried letting people add “smart straps” to augment their watches after the fact. I feel like UNA follows that legacy, except that the modules will slot directly onto the watch itself. The question will be how many modules you’ll be able to add.

The UNA Watch Kickstarter will take place until May 2, with plenty of early-bird deals available to save $75 on the eventual list price. It’s expected to ship in August 2025. Meanwhile, if you’re more interested in Pebble, its new watches are also available for preorder now, with lower prices but a later scheduled release date of December 2025.

El Salvador has introduced a new artificial intelligence (AI) law to foster innovation while ensuring responsible development and regulatory oversight.

On Feb. 28, the country’s National Bitcoin Office confirmed the legislation’s approval, highlighting its role in integrating AI into key sectors while maintaining data security and ethical standards.

This new legal framework establishes clear guidelines for AI research and application, making it the first national law to prioritize open-source AI development. It builds on existing regulations, such as the Personal Data Protection Act and the Cybersecurity and Information Security Act, to create a structured environment for AI advancement.

However, the AI law specifies that in the event of any conflict, the AI Law will take precedence over the Personal Data Protection Law. In other words, while the AI law works alongside data protection legislation, it will override the Data Protection Act where necessary to facilitate AI development.

The Nayib Bukele-led government sees this initiative as a strategic step toward attracting investment and technological growth.

The country’s lawmakers believe that by embracing AI, the government can unlock new economic opportunities, particularly for young professionals eager to contribute to the evolving digital landscape.

El Salvador’s AI law

The law provides legal certainty for AI developers, ensuring protection from early-stage experimentation to full-scale implementation. It also introduces safeguards against the misuse of AI technologies while addressing proprietary and open-source models.

A key focus of the legislation is promoting AI research by establishing dedicated laboratories and innovation centers.

These institutions will explore AI-driven solutions for public services, including traffic management, water resource monitoring, and geothermal energy optimization. The goal is to enhance the country’s infrastructure and improve the quality of life for its citizens.

Additionally, the framework prioritizes education and workforce training by equipping institutions with state-of-the-art AI resources. This initiative aims to build a strong domestic AI talent pool, positioning El Salvador as an AI research and development hub.

Meanwhile, the law mandates the creation of the National Agency for Artificial Intelligence (ANIA) to ensure compliance with ethical standards. This agency will oversee AI regulation and enforce guidelines to prevent misuse.

ANIA will also manage the National Registry of Artificial Intelligence Development, a system designed to track AI developers and ensure transparency in the sector.

The open-source model, OpenThinker-32B, achieved similar or better results in key performance tests while requiring far less training data, according to the February 12 blog post.

OpenThinker-32B was trained using just 114,000 examples, a fraction of DeepSeek’s 800,000. The dataset, called OpenThoughts-114k, included detailed solutions, coding test cases, starter code, and subject-specific information.

Did you know?

Want to get smarter & wealthier with crypto?

Subscribe – We publish new crypto explainer videos every week!

Crypto Mining Explained: How to Earn From Mining Bitcoin? (Animated)

Training took about 90 hours on four nodes, each equipped with eight H100 GPUs. Another dataset, containing 137,000 unverified samples, was processed separately using Italy’s Leonardo Supercomputer, consuming 11,520 A100 GPU hours in just 30 hours.

When tested, OpenThinker-32B delivered a 90.6% accuracy rate on the MATH500 benchmark, surpassing DeepSeek’s 89.4%. It also scored 61.6 on GPQA-Diamond, compared to DeepSeek’s 57.6, which shows strength in general reasoning tasks.

For coding tasks, OpenThinker-32B lagged slightly behind, scoring 68.9 against DeepSeek’s 71.2. Since OpenThinker-32B is open source, these numbers could improve as developers contribute refinements.

Built on Alibaba’s Qwen2.5-32B-Instruct language model, OpenThinker-32B supports a 16,000-token context window. While this is smaller than other AI models, it is still enough to handle complex equations and long programming tasks.

On February 13, Elon Musk revealed the latest version of xAI’s chatbot, Grok 3. What can it do? Read the full story.