Non-US LLMs: what are the main AI models not from the United States?

by Stélio Inácio, Founder at Jon AI and AI Specialist

The Global AI Stage: A World of Competitors

While news headlines often focus on AI from American companies, a dynamic and powerful AI ecosystem is thriving across the globe. Nations are pursuing distinct strategies to shape the future of artificial intelligence, driven by unique economic goals, cultural values, and geopolitical ambitions. From China's state-directed mission for technological supremacy to Europe's push for a regulated, rights-based AI, the global landscape is anything but uniform. This lesson explores the major AI models and strategies emerging from key international players.

China: The State-Driven Superpower

China's approach to AI is defined by a powerful top-down national strategy. Through initiatives like "Made in China 2025" and the "AI+ initiative," the government is driving for technological self-sufficiency and the deep integration of AI across its entire economy, from manufacturing to healthcare. This state-led push is not just about funding; the government also fosters intense domestic competition by backing multiple startups—often called the "Six Tigers"—to create "battle-hardened" national champions capable of competing globally.

Key Chinese AI Models

Ernie (Baidu): Developed by the search giant Baidu, the Ernie series of models are powerful and multimodal, with a strong focus on "deep-thinking reasoning" capabilities.
Qwen (Alibaba): Alibaba's Qwen family is known for its technical sophistication, featuring a range of models from tiny, efficient ones to massive Mixture-of-Experts (MoE) models with hundreds of billions of parameters.
Hunyuan (Tencent): Tencent's Hunyuan models are designed for large-scale deployment and are deeply integrated into its 700+ products. The family includes massive MoE models and even a first-of-its-kind open-source 3D asset generator.
GLM (Zhipu AI): Emerging from Tsinghua University, Zhipu AI's GLM series is a flagship of China's startup scene, with its creators claiming its GLM-4-Plus model has achieved parity with GPT-4.
Kimi (Moonshot AI): This chatbot has gained significant popularity in China for its remarkable ability to process up to two million Chinese characters in a single prompt.

Europe: A Continent of Sovereign Ambitions and Regulation

Europe's AI landscape is a fascinating mix of distinct national strategies united under a single, powerful regulatory framework: the EU AI Act. This act, the world's first comprehensive AI law, establishes a risk-based approach, prioritizing safety, transparency, and fundamental rights. This regulation-first philosophy shapes the environment for key players like France and Germany.

Key European AI Models

Mistral (France): Based in Paris, Mistral AI has rapidly become a global open-source champion. Its strategy is to release powerful and efficient models under permissive licenses, which has earned it a massive global following and positioned it as a key challenger to proprietary US models.
Luminous & Pharia (Aleph Alpha, Germany): This German startup has carved out a niche by focusing on "trustworthy" and "explainable" AI for enterprise and government clients. Its models are designed to comply with strict European data laws and are trained on five European languages, emphasizing the goal of technological sovereignty.

Concept Spotlight: Open Source as a Strategic Tool

For many non-US players, competing with the immense financial and data resources of American tech giants is difficult. They have turned to open-source as a powerful "asymmetric competitive weapon." By freely releasing their model's code and weights, companies like France's Mistral and the UAE's TII achieve several strategic goals:

They build a loyal global community of developers who use and improve their technology.
They lower the barrier to entry, allowing smaller companies to build on their work.
They foster trust through transparency, which is a key advantage in the privacy-conscious European market.
They enable regional initiatives like OpenEuroLLM and SEA-LION to create models that preserve local languages and cultures, directly countering the dominance of English-centric US models.

Other Global Hubs: Diverse Strategies and Niche Strengths

Beyond China and Europe, other nations are carving out influential roles by focusing on their unique strengths.

United Arab Emirates (UAE): The UAE is pursuing an ambitious "investment-led infrastructure model." Its strategy is to use its vast sovereign wealth to control the "new oil"—computing power. Through its national champion G42 and projects like the "Stargate" supercomputing cluster, the UAE aims to become a global hub for AI infrastructure. At the same time, its Technology Innovation Institute (TII) develops the powerful open-source Falcon LLM.
Canada: Canada has adopted a "research-first" approach, focusing on cultivating a world-class talent pipeline through institutions like Mila in Montreal. This has made it a "Switzerland of AI"—a neutral, talent-rich hub that attracts global collaboration and investment. This ecosystem has produced leading companies like Cohere.
Other Key Players:
- Israel is a powerhouse for AI startups, particularly in cybersecurity and defense.
- Japan is focusing on robotics and automation to address its demographic challenges.
- South Korea leverages its hardware dominance to advance AI in smart cities and autonomous vehicles.
- India is rapidly growing its AI talent pool and fostering a vibrant startup ecosystem.

Quick Check

What is a key characteristic of China's national strategy for developing AI?

A) A hands-off, market-led approach focused on attracting foreign investment.

B) A top-down, state-driven mission to achieve technological self-reliance and global leadership.

C) A primary focus on creating strict, rights-based regulations before developing technology.

The UAE's national AI strategy is primarily focused on what?

A) Creating the world's strictest, rights-based regulations for AI safety.

B) Fostering a "pro-innovation" market by attracting as much foreign capital as possible.

C) Leveraging sovereign wealth to build massive-scale AI compute infrastructure and controlling it as a strategic resource.

Recap: The World of Non-US AI

What we covered:

The AI world is multipolar, with many countries developing powerful models and unique national strategies.
China's state-driven approach aims for technological self-reliance with models like Ernie and Qwen.
Europe is defined by its regulation-first approach and a push for "sovereign AI" with champions like France's Mistral and Germany's Aleph Alpha.
The UAE is using its wealth to become an infrastructure superpower, controlling compute as the "new oil."
Canada has focused on a research-first strategy to become a global hub for AI talent.
Open-source has become a key competitive strategy for many non-US players to challenge the dominance of proprietary American models.

Why it matters:

Understanding the global AI landscape shows that innovation is happening everywhere. The values and goals of the country where an AI is built—be it a focus on state control, individual rights, or economic power—can be reflected in the technology itself.

Next up:

We'll dive into the concept of multi-modal AI, exploring how models are learning to understand not just text, but images, voice, and video all at once.

Jon AI Services