Qwen-SEA-LION-v4-32B-IT
Qwen-SEA-LION-v4-32B-IT is a text model derived from Qwen3-32B. AI Singapore continues pretraining on about 100 billion tokens from SEA-Pile v2, then post-trains on about eight million question-and-answer pairs.
How this release works
The checkpoint card specifies a native context length of 32,768 tokens. Its language list includes Burmese, English, Indonesian, Khmer, Lao, Malay, Mandarin, Tagalog, Tamil, Thai and Vietnamese. The continued-training stage distinguishes it from releases adapted only with instruction examples.Qwen-SEA-LION-v4-32B-IT official model card
How to assess it for your work
Practical guidance. Choose this release as a text baseline when you want to assess the effect of regional pretraining. Keep document length within the native limit during that comparison. Do not infer a larger verified limit from a family overview or from the capabilities of a different Qwen release.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Model identifiers | aisingapore/Qwen-SEA-LION-v4-32B-IT |
| Context window | 32,768 tokens Native context stated in the individual checkpoint card. Qwen-SEA-LION-v4-32B-IT official model card |
| Base model | Qwen/Qwen3-32BQwen-SEA-LION-v4-32B-IT official model card |
| Model size | 32 billion parametersQwen-SEA-LION-v4-32B-IT official model card |
| License declared by the model card | MITQwen-SEA-LION-v4-32B-IT official model card |
| Input and output | TextQwen-SEA-LION-v4-32B-IT official model card |
Related models
- Nemotron-SEA-LION-v4.8-30B-A3B
AI Singapore · SEA-LION v4.8
A regional instruction model that combines Mamba2 and Transformer layers with mixture-of-experts routing.
Launch date not confirmedDownloadable - Qwen-SEA-LION-v4.5-27B-IT
AI Singapore · SEA-LION v4.5
A 27B multimodal model based on Qwen3.6, distilled for Southeast Asian language tasks.
Launch date not confirmedDownloadable