Blog
-
Announcing Qwen-SEA-LION-v4 (4B & 8B): Additional Vision-Language Models for Southeast Asia
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…
-
SEA-LION Summit 2025: Powering Southeast Asia’s AI Future
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,…
-
Announcing Qwen-SEA-LION-v4: Advanced Reasoning and Language Depth for Southeast Asia
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…
-
Bridging Cultures and Pixels: Introducing SEA-LION-VL, a New Vision-Text Model for Southeast Asia
We are thrilled to announce the launch of Gemma-SEA-LION-v4-27B-VL, a cutting-edge, instruction-tuned vision-text model designed specifically for the unique and diverse landscape of Southeast Asia. This release is the result of a powerful collaboration between the Products Pillar at AI Singapore and SEACrowd, with funding from the Singapore National Research Foundation (NRF). A Multimodal Model…
-
Introducing SEA-Guard: A Specialized Safety Model for AI in Southeast Asia
As large language models (LLMs) become part of our everyday digital lives, ensuring their outputs are safe, ethical, and compliant is more important than ever. That’s where SEA-Guard comes in — a specialized model built to act as a protective layer around foundation models, keeping both user inputs and AI outputs within safe boundaries. What…
-
5th Languages Summit: Southeast Asia and India Power Up to Supercharge AI
On 22 July 2025, the 5th Languages Summit took place in Bangalore, India, jointly hosted by AI Singapore and Google. The event at Google Ananta brought together a vibrant community of innovators to shape the next chapter of AI in Southeast Asia and India. Overview of the Summit The 5th Languages Summit wasn’t just another…
-
SEA-LION v4: Our First Multimodal Release, Open, Powerful & Small
We’re proud to introduce SEA-LION v4 – our open, most powerful and efficient, multimodal & multilingual model yet, designed to run even on a laptop. This release expands SEA-LION beyond text to support image + text inputs, while staying true to our focus on Southeast Asian languages, culture, and use cases. Built on Gemma 3…
-
Release of WangchanLION v3
WangchanLION-v3: An Open-Science of Large-scale Thai Pre-training Using SEA-LION Introduction We are very happy to announce the release of WangchanLION-v3, with the collaboration of AI Singapore, VISTEC, and SCB10X. WangchanLION-v3 is an 8 billion parameter model, pre-trained on 47 billion high-quality Thai tokens. The 47B tokens will also be released on AI Singapore’s HuggingFace. While…
-
AI Town: Multilingual AI Simulations with SEA-LION
AI Town presents a fascinating microcosm, a virtual environment where AI-driven characters live, chat and socialise. South East Asian Languages in One Network (SEA-LION) is a family of open-source Large Language Models (LLMs) that better understands Southeast Asia’s (SEA) diverse contexts, languages, and cultures. This post describes the installation of AI Town, and how SEA-LION…
-
4th Languages Summit: Bringing Southeast Asia’s Languages to the forefront of AI
On 23 April 2025, AI Singapore and Google co-hosted the fourth Language Summit at Google Singapore, bringing together over 80 policymakers, researchers, industry experts, and community leaders from across Southeast Asia and beyond. Unveiling Project Aquarium At this summit, we were excited to unveil Aquarium (Beta), a collaborative brainchild of AI Singapore, Google, and Project…
-
Introducing Aquarium: An Open Data Platform for Southeast Asian Languages
Southeast Asia is one of the most linguistically diverse regions in the world, with over 650 million people speaking hundreds of languages and dialects. Indeed, many Southeast Asian languages are still missing or underrepresented in the data used to train today’s most powerful AI models. That is a problem – AI that doesn’t understand us…
-
SEA-LION v3.5 and Updated v3: Enhanced Language Models for Southeast Asia
We are proud to launch SEA-LION v3.5, our first set of hybrid reasoning models trained on Southeast Asian data. Mode selection is managed through the tokenizer’s chat template and offers versatile functionality, handling both complex reasoning tasks and general text generation. Trained off our SEA-LION v3, SEA-LION v3.5 is explicitly enhanced for reasoning tasks with…
-
Assessing LLM Performance for Southeast Asia: Introducing SEA-HELM
In collaboration with the HELM team at Stanford CRFM (with special thanks to Yifan Mai), we would like to announce the official release of SEA-HELM (Southeast Asian Holistic Evaluation of Language Models), a general-purpose holistic benchmark to evaluate the performance of LLMs in the Southeast Asian context. Distinctive Features Cutting-edge Components Counts for Selected Tasks…
-
OpenAI-compatible APIs with SEA-LION and Bedrock Access Gateway
In our earlier post, we have shown how to leverage on Amazon Bedrock‘s Custom Model Import to deploy SEA-LION on the cloud. After importing the SEA-LION models, you can build applications with the AWS SDK. This article describes an alternative method to build the applications with OpenAI-compatible APIs served by the Bedrock Access Gateway. Prerequisites…
-
Importing and Using SEA-LION in a Serverless, On-Demand Environment with Amazon Bedrock
The SEA-LION models are open source and freely available for research and commercial use. A common question we receive from developers is how to host and configure SEA-LION for model inference in their own environments. When it comes to deploying our models, organizations have several hosting options available to them. One such approach involves using…
-
Towards fair and comprehensive multilingual LLM benchmarking
Explore how to design fair, transparent, and representative multilingual evaluations for large language models. Large language models (LLMs) can now generate complete and coherent sentences in many more languages than most humans. Emergent abilities, such as reasoning, creative writing, and human-agent interactions, have propelled enterprises to adapt these models to suit a wide range of…
-
SEA-LION v3: 128K Context Length and 70B Models
We are excited to announce the release of two new variants for SEA-LION v3, our latest large language models tailored specifically for Southeast Asian languages. Building upon Meta’s Llama and SEA-LION’s data, these variants have strong capabilities in handling diverse linguistic and cultural nuances inherent to Southeast Asian region. New SEA-LION v3 Variants 1. SEA-LION…
-
Building a Multilingual Chatbot with SEA-LION: A Step-by-Step Guide
In today’s rapidly evolving digital landscape, the ability to communicate across languages is more important than ever. Whether you’re building a customer service bot, an AI assistant, or any other type of interactive application, ensuring your chatbot can handle multiple languages is a step toward reaching a global audience. Enter SEA-LION — a multilingual large…
-
SEA-LION v3: Advancing Southeast Asian AI in An Open, Inclusive, and Responsible Manner
The latest evolution in Southeast Asian AI, built with assistance from regional collaborators, and global partners such as Google and Nvidia. Introducing SEA-LION v3 Continued Pre-training The latest in our family of open-source models, SEA-LION v3, was built by continued pre-training of Gemma2 9B on 200 billion high quality Southeast Asian tokens from 11 official…
-
Announcing Qwen-SEA-LION-v4 (4B & 8B): Additional Vision-Language Models for Southeast Asia
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…
-
Assessing LLM Performance for Southeast Asia: Introducing SEA-HELM
In collaboration with the HELM team at Stanford CRFM (with special thanks to Yifan Mai), we would like to announce the official release of SEA-HELM (Southeast Asian Holistic Evaluation of Language Models), a general-purpose holistic benchmark to evaluate the performance of LLMs in the Southeast Asian context. Distinctive Features Cutting-edge Components Counts for Selected Tasks…
-
Towards fair and comprehensive multilingual LLM benchmarking
Explore how to design fair, transparent, and representative multilingual evaluations for large language models. Large language models (LLMs) can now generate complete and coherent sentences in many more languages than most humans. Emergent abilities, such as reasoning, creative writing, and human-agent interactions, have propelled enterprises to adapt these models to suit a wide range of…
-
AI Town: Multilingual AI Simulations with SEA-LION
AI Town presents a fascinating microcosm, a virtual environment where AI-driven characters live, chat and socialise. South East Asian Languages in One Network (SEA-LION) is a family of open-source Large Language Models (LLMs) that better understands Southeast Asia’s (SEA) diverse contexts, languages, and cultures. This post describes the installation of AI Town, and how SEA-LION…
-
OpenAI-compatible APIs with SEA-LION and Bedrock Access Gateway
In our earlier post, we have shown how to leverage on Amazon Bedrock‘s Custom Model Import to deploy SEA-LION on the cloud. After importing the SEA-LION models, you can build applications with the AWS SDK. This article describes an alternative method to build the applications with OpenAI-compatible APIs served by the Bedrock Access Gateway. Prerequisites…
-
Importing and Using SEA-LION in a Serverless, On-Demand Environment with Amazon Bedrock
The SEA-LION models are open source and freely available for research and commercial use. A common question we receive from developers is how to host and configure SEA-LION for model inference in their own environments. When it comes to deploying our models, organizations have several hosting options available to them. One such approach involves using…
-
Building a Multilingual Chatbot with SEA-LION: A Step-by-Step Guide
In today’s rapidly evolving digital landscape, the ability to communicate across languages is more important than ever. Whether you’re building a customer service bot, an AI assistant, or any other type of interactive application, ensuring your chatbot can handle multiple languages is a step toward reaching a global audience. Enter SEA-LION — a multilingual large…