Typhoon2.5-Qwen3-4B: A Thai-Centric LLM with Function Calling

Typhoon2.5-Qwen3-4B, developed by SCB 10X, is a 4 billion parameter instruction-tuned large language model built upon the Qwen3 architecture. It stands out with its impressive 256K context length, enabling it to process and understand very long inputs and conversations. The model is primarily optimized for the Thai language but also supports English, making it a versatile tool for bilingual applications.

Key Capabilities

• Bilingual Support: Excels in both Thai and English language processing.
• Extended Context Window: Features a 256K token context length, ideal for complex, multi-turn interactions or analyzing lengthy documents.
• Function Calling: Capable of understanding and executing tool calls, allowing integration with external systems and APIs for enhanced functionality.
• Instruction Following: Designed to follow instructions effectively, making it suitable for various AI assistant roles.

Intended Use Cases

• Thai Language Applications: Ideal for chatbots, content generation, and analysis in Thai.
• Bilingual Assistants: Can serve as an AI assistant handling queries and tasks in both Thai and English.
• Tool-Augmented Systems: Suitable for scenarios requiring interaction with external tools or APIs through its function-calling feature.
• Research and Development: Provides a strong base for further fine-tuning and experimentation, particularly in long-context and multilingual NLP tasks.