Machine Translation (MT) has come a long way in the era of Large Language Models (LLMs). Scores on translation benchmarks continue to rise and we are starting to see improvements even in the lower-resourced languages of our Southeast Asian (SEA) region. Despite this encouraging trend, we still observe specific issues, both in our own evaluations…
When you speak to an AI agent instead of typing, how much intelligence is actually lost along the way? While Large Language Models (LLMs) continue to post impressive scores on text benchmarks, real-world deployment is increasingly shifting towards voice. No matter how speech reaches the model, the task it solves is the one it heard,…
Advancing AI Safety with Action Governance, Multi-Turn Evaluation, and Custom Policy Enforcement – Jev-style Prediction, Powered by Gemma4 Models As AI agents transition from simple conversational chatbots into autonomous tools operating in complex, multi-turn environments, safety guardrails must evolve alongside them. Traditional moderation models—designed primarily for single-turn safety classification—frequently fail when confronted with subtle regional…
On 15 October 2026, AI Singapore and Google co-hosted the 6th ATLAS Language Summit at Google Singapore. It brought together 189 participants: 107 in the room and 82 online, from 70 organisations in 24 countries across Southeast Asia, Asia, Africa and beyond – more than doubling the representation from our previous Language Summit in 2025….
AI Singapore is thrilled to announce the Nemotron-SEA-LION-v4.8 suite of models, developed in deep technical collaboration with NVIDIA. By combining NVIDIA NeMo Megatron Bridge and NVIDIA NeMo AutoModel training frameworks with AI Singapore’s extensive regional data expertise, the partnership has produced open-weights models built specifically for Southeast Asia’s diverse linguistic landscape. Built on the NVIDIA…
We are glad to introduce the SEA-LION v4.5 family, which adds state-of-the-art models that work well as agents but also understand Southeast Asian (SEA) languages, cultures, and local knowledge to the SEA-LION family of models. SEA-LION v4.5 models benefit from new post-training techniques: knowledge distillation, targeted supervised fine-tuning, and smarter model merging. With these methods,…
AI is poised to transform the world, and to ensure this transformation benefits everyone, it must be inclusive of all languages and cultures. This is particularly crucial in regions like Southeast Asia, where a wealth of local languages and deep cultural nuances are often overlooked by today’s global Large Language Models (LLMs) despite being home…
Vision-Language Models (VLMs) are changing how AI perceives and describes the world—but does your AI see and explain the real Southeast Asia? Adding to our family of models from Llama, Gemma and Qwen, we are proud to announce the release of small but powerful multilingual and multicultural text models. Like Gemma, these models also support…
AISG brings together developers, industry leaders, and regional partners to showcase how open, localised AI is solving real-world challenges across Southeast Asia. AI Singapore (AISG) hosted the inaugural SEA-LION Summit: Powering Southeast Asia’s AI Future at AIMX Singapore, bringing together policymakers, developers, researchers, and organisations from across the region for an afternoon of knowledge sharing,…
We are proud to introduce Qwen-SEA-LION-v4-32B-IT, a new iteration of our powerful multilingual model series specifically trained for Southeast Asia. This initial experimental version represents a significant architectural evolution, transitioning from a sentence-piece tokenizer to a more modern byte-pair encoding (BPE) for superior text handling. Built upon the powerful Qwen3-32B foundation model, Qwen-SEA-LION-v4-32B-IT inherits long-context…