Mistral’s Singapore office will also serve as an extended science hub where the company will develop frontier models for the region’s public sector, high-tech manufacturing, and robotics industries, according to the company.
Chinese open-weight models, which are free to download and run on users’ own hardware, are already widely used in Southeast Asia. Singapore’s national AI R&D program built its SEA-LION model on Alibaba’s Qwen, and OCBC runs more than 30 internal banking tools on open models, including DeepSeek.
Mistral says it can compete by offering companies hands-on support. Marjorie Janiewicz, the company’s chief revenue officer, described three components to delivering an AI model: the intelligence, the technology to deploy it, the technology to turn it into workflows and applications, and forward-deployed engineering. She said that one or more of those components are usually missing in Chinese AI models when companies try to get them to production.
Geoff Soon, Mistral’s vice president of revenue for Asia-Pacific, said more than 70 per cent of the Singapore team are forward-deployed engineers and scientists, adding that the company is focused on building a deep scientific core in Singapore, with recent hires in science and engineering.
The company’s ties to Singapore date back to 2024. Soon recalled that as a 20-person company founded only six months earlier, Mistral received an outreach from the Singapore Ministry of Defence that initially went to spam but was followed by a phone call asking them to take it seriously. That led to a partnership with MINDEF.
In November 2025, Mistral signed an agreement with Singapore’s Home Team Science and Technology Agency, a statutory board under the Ministry of Home Affairs, to collaborate on video AI, embodied AI, and AI for cybersecurity.
Janiewicz argued that companies do not have to rely on closed-source model providers or choose only between the United States and China, positioning Mistral as an alternative for global companies seeking autonomy.
The company benefits from geopolitical tailwinds as European countries increasingly discuss developing sovereign AI models not tied to American or Chinese technology. U.S. Commerce Department restrictions on foreign access prompted Anthropic to suspend two of its most advanced AI models for all users worldwide on June 12. Mistral’s executives said that episode turned sovereignty from a talking point into a procurement requirement. Janiewicz said sovereignty was now a reality, with organizations looking for business continuity in the partners they select to build their AI systems with.
Mistral’s most recent model, Mistral Large 4, features 1 trillion total parameters and a 1-million-token context window, and outperforms other open-weight models like DeepSeek Flash V4.1 and GLM 5.3 on the Coding Agent Index according to Artificial Analysis.
