Qwen-SEA-LION-v4.5-27B-IT-GGUF

Last update: 2026-06-18

This repository contains the GGUF quantized weights for Qwen-SEA-LION-v4.5-27B-IT, our flagship 27-billion parameter instruction-tuned model built upon the state-of-the-art Qwen3.6-27B architecture.

The model is heavily optimized for Southeast Asian (SEA) languages, cultures, and contexts. It has undergone rigorous post-training—consisting of knowledge distillation, supervised fine-tuning (SFT) on curated regional datasets, and reinforcement learning (RL). This model brings high-capacity reasoning, tool-use, and advanced agentic workflows to local server environments and high-end consumer hardware with minimized memory requirements.

Model Details

Model Description

SEA-LION stands for Southeast Asian Languages In One Network.

We performed post-training in English and SEA languages on Qwen3.6-27B, a multimodal learning model using the Qwen3.6 architecture, to create Qwen-SEA-LION-v4.5-27B-IT.

For tokenization, the model employs the default tokenizer used in Qwen3.6.

• Developed by: AI Products Pillar, AI Singapore
• Funded by: Singapore NRF
• Shared by: AI Products Pillar, AI Singapore
• Model type: Causal Language Model with Vision Encoder
• Training Stage: Post-Training (Logit Distillation & Model Merging))
• Context length: 262k
• Language(s): fine-tuned on Burmese, Indonesian, Filipino, Malay, Tamil, Thai, and Vietnamese
• License: MIT
• Finetuned from model: aisingapore/Qwen-SEA-LION-v4.5-27B-IT

Available Quantized Versions

We provide multiple quantization formats to optimize deployment trade-offs between memory footprint and output quality.

File / Quantization	File Size
Q4_K_M	16.8 GB
Q6_K	22.4 GB
Q8_0	29 GB
F16	54.7 GB

Model Strengths

• High-Capacity Foundation: Derived from the Qwen3.6-27B architecture, delivering advanced multi-turn reasoning, complex logical planning, and code generation.
• Native Context Window (262K): Supports long-context parsing, extensive document retrieval (RAG), and persistent history retention in complex multi-turn chats.
• Broad Linguistic Coverage: Fine-tuned on native language data covering Burmese, Indonesian, Filipino (Tagalog), Malay, Tamil, Thai, and Vietnamese, alongside competitive English and Chinese capabilities.
• Advanced Agentic Capabilities: Highly precise tool-use, schema validation, and nested function-calling, backed by specialized reinforcement learning checkpoints.

How to Get Started

Using llama.cpp

You can serve the model using llama.cpp (add other flags to customize your setup):

llama-server -ngl -1 \
    --mmproj /directory/path/to/mmproj-Qwen-SEA-LION-v4.5-27B-IT-F16.gguf \
    -m /directory/path/to/Qwen-SEA-LION-v4.5-27B-IT-Q4_K_M.gguf

You can proceed to make API calls to the model:

curl -X POST http://localhost:8080/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "Qwen-SEA-LION-v4.5-27B-IT-Q4_K_M.gguf",
    "messages": [
        {
        "role": "user",
        "content": "Teach me a greeting in Bahasa Indonesia."
        }
      ],
    "chat_template_kwargs": {
      "enable_thinking": false
      }
  }'

You can also access the UI at http://localhost:8080 to chat with the model directly.

Uses

Out-of-Scope Use

The model has not been aligned for safety. Developers and users should perform their own safety fine-tuning and related security measures. In no event shall the authors be held liable for any claims, damages, or other liabilities arising from the use of the released weights and codes.

Bias, Risks, and Limitations

The model was not tested for robustness against adversarial prompting. It is important for users to be aware that our model exhibits certain limitations that warrant consideration. Like many LLMs, the model can hallucinate and occasionally generates irrelevant content, introducing fictional elements that are not grounded in the provided context. Users should also exercise caution in interpreting and validating the model's responses due to the potential inconsistencies.

More Information

This is the repository for the commercial instruction-tuned model. The model has not been aligned for safety. Developers and users should perform their own safety fine-tuning and related security measures. In no event shall the authors be held liable for any claims, damages, or other liabilities arising from the use of the released weights and codes.

For more info, please contact us at sealion@aisingapore.org

Acknowledgement

This project is supported by the National Research Foundation Singapore and Infocomm Media Development Authority (IMDA), Singapore under its National Large Language Model Funding Initiative.

Contact

sealion@aisingapore.org