Home > Models & Research

Models & Research

Pioneering the Future

MERaLiON (Multimodal Empathetic Reasoning and Learning in One Network) is Southeast Asia’s empathetic Multimodal Large Language Model (MLLM), designed to understand the region’s diverse languages, cultures, and communication styles.

MERaLiON Models

Our research is organized into core collections representing the evolution of Southeast Asian-centric AI. We prioritise transparency by releasing our model weights and benchmarks for community evaluation. Explore all MERaLiON models and resources on Hugging Face.

Collection	Versions
MERaLiON-3 (Preview)	10B-preview
MERaLiON-2	10B, 10B-ASR, 10B-MLX, 3B, 3B-MLX
Speech Emotion Recognition	SER v1
Speech-Encoder	SpeechEncoder-v1, SpeechEncoder-2
MERaLiON-1	AudioLLM-Whisper-SEA-LION

Build with MERaLiON

Playground

Experience MERaLiON. Upload or
record an audio clip, then ask
anything: transcribe, translate,
summarise, detect emotions.

Try the MERaLiON Playground

API Console

Access MERaLiON via the API
console. Ideal for developers
building applications.

Get MERaLiON API Access

Download

Download and deploy MERaLiON
models directly on your own systems.

Download & Self-Host

Connect

Connect with the team to discuss
consortium membership and co-
development opportunities.

Join the Consortium

Research Library

No.	List of Papers	Date
1.	AdaMCoT: Rethinking Cross-Lingual Factual Reasoning through Adaptive Chain-of-Thought	27 Jan 2026
2.	Latent-RQ: Enhancing Speech Pre-training with Latent Representations and Random Quantization	27 Jan 2026
3.	Train Multi-Modal LLM to Understand Diverse Speech Paralinguistics by Distilling from Teacher with Meta-Information Prompt	27 Jan 2026
4.	IFEval-Audio: Benchmarking Instruction-Following Capability in Audio-based Large Language Models	12 Nov 2025
5.	MERaLiON-SER: Robust Speech Emotion Recognition Model for English and SEA Languages	7 Nov 2025
6.	A Benchmark for Translations Across Styles and Language Variant	4 Nov 2025
7.	Beyond Classification: Towards Speech Emotion Reasoning with Multitask AudioLLMs	29 Sep 2025
8.	Benchmarking Contextual and Paralinguistic Reasoning in Speech-LLMs: A Case Study with In-the-Wild Data	24 Sep 2025
9.	Incorporating Contextual Paralinguistic Understanding in Large Speech-Language Models	10 Aug 2025
10.	MERaLiON-AudioLLM: Advancing Speech and Language Understanding for Singapore	27 Jul 2025
11.	CCL-XCoT: An Efficient Cross-Lingual Knowledge Transfer Method for Mitigating Hallucination Generation	17 Jul 2025
12.	Contextual Paralinguistic Data Creation for Multi-Modal Speech-LLM: Data Condensation and Spoken QA Generation	19 May 2025
13.	Advancing Singlish Understanding: Bridging the Gap with Datasets and Multimodal Models	2 Jan 2025
14.	MERaLiON-SpeechEncoder: Towards a SpeechFoundation Model for Singapore and Beyond	20 Dec 2024
15.	MERaLiON-AudioLLM: Technical Report	13 Dec 2024
16.	MoWE-Audio: Multitask AudioLLMs with Mixture of Weak Encoders	10 Sep 2024
17.	PRESENT: Zero-Shot Text-to-Prosody Control	13 Aug 2024
18.	AudioBench: A Universal Benchmark for Audio Large Language Models	23 Jan 2024
19.	SeaEval for Multilingual Foundation Models: From Cross-Lingual Alignment to Cultural Reasoning	9 Sep 2023

Acknowledgements

Developed by A*STAR Institute for Infocomm Research (A*STAR I²R), MERaLiON is one of two national LLMs under Singapore’s S$70M National Multimodal Large Language Model Programme (NMLP), supported by the National Research Foundation (NRF) and the Infocomm Media Development Authority (IMDA). First launched in December 2024, MERaLiON is an empathetic and culturally attuned Artificial Intelligence (AI) model designed to support natural interactions across sectors, reflecting Singapore’s ambition to lead in human-centric, regionally grounded AI. The high-performance computing (HPC) resources used to train the MERaLiON model were provided by the National Supercomputing Centre (NSCC) Singapore through its ASPIRE 2A+ supercomputer.