Qwen-SEA-LION-v4-4B-VL is a vision-language model with 4 billion parameters built on the Qwen3-VL-4B-Instruct architecture. It has been specifically instruction fine-tuned for the Southeast Asian region, possessing multilingual and multicultural capabilities, supporting English and seven Southeast Asian languages, and retaining strong vision-language understanding capabilities.

MultimodalTransformersMultiple Languages

Qwen-SEA-LION-v4-32B-IT-4BIT is a 4-bit quantized version of the 32B parameter large language model optimized for Southeast Asian languages. It significantly reduces memory requirements while maintaining excellent performance and can run on consumer-grade hardware.

Natural Language ProcessingTransformersMultiple Languages

Qwen-SEA-LION-v4-32B-IT is a large Southeast Asian language model built on Qwen3-32B, specifically pre-trained and instruction fine-tuned for the Southeast Asian region. This model continues pre-training on the SEA-Pile v2 corpus containing 7 Southeast Asian languages and undergoes instruction fine-tuning on 8 million pairs of high-quality question-and-answer data, possessing strong multilingual understanding and reasoning abilities.

Natural Language ProcessingTransformersMultiple Languages