Cloudflare has announced the expansion of its regional artificial intelligence (AI) model offerings, enabling developers to build with models that support their own digital sovereignty. With this move, Cloudflare said it aims to empower governments and enterprises to deploy AI tailored to local languages, regulations, and values.
The company highlighted that while some AI technologies are designed to work on-device, many require additional power for inference tasks. Cloudflare’s distributed network provides a platform to run these workloads closer to users, while its Workers AI and AI Gateway services ensure reliable, scalable, and cost-efficient deployment.
As part of the announcement, Cloudflare introduced region-specific AI models:
- India: AI4Bharat’s IndicTrans2, an open-source language model supporting translations across 22 Indian languages.
- Japan: Preferred Networks, Inc. (PFN)’s PLaMo-Embedding-1B, a home-grown Japanese text embedding model made freely and openly available.
- Southeast Asia: SEA-LION v4-27B, a bilingual large language model developed with Singapore’s National Supercomputing Centre and AI Singapore, supporting multiple regional languages.
Cloudflare noted that open-sourcing these large language models (LLMs) reflects the recognition that AI sovereignty is not an end in itself but a means to unlock broader innovation and inclusion. Countries including Singapore, India, and Japan have chosen to open-source many of their local models, ensuring accessibility to the global developer community.
Through its Connectivity Cloud, Cloudflare protects entire corporate networks, helping businesses build securely online. The company said its latest AI initiatives further extend its vision of making the Internet more secure, reliable, and efficient for everyone.