Gemma-SEA-LION-v4.5-E2B-IT
Gemma-SEA-LION-v4.5-E2B-IT adapts Google's Gemma 4 E2B instruction model through distillation and merging. It supports text, image, video and audio inputs. Its stated size is 2.3 billion effective parameters, or 5.1 billion when embeddings are included.
How this release works
The regional training uses SEA-Instruct-2602. The card describes configurable thinking, system-role support and function calling. Its listed regional languages are Burmese, Indonesian, Filipino, Malay, Tamil, Thai and Vietnamese, alongside English. Multimodal input support does not imply generated speech or video.Gemma-SEA-LION-v4.5-E2B-IT official model card
How to assess it for your work
Practical guidance. Consider this release when a local assistant must read a picture as well as answer a message. Verify each required input type through the exact runtime you plan to ship. Keep the full model memory requirement separate from the effective-parameter label, especially when choosing a mobile or laptop deployment.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Model identifiers | aisingapore/Gemma-SEA-LION-v4.5-E2B-IT |
| Base model | google/gemma-4-E2B-itGemma-SEA-LION-v4.5-E2B-IT official model card |
| Distillation teacher | google/gemma-4-31B-itGemma-SEA-LION-v4.5-E2B-IT official model card |
| Context stated by the card | 128K tokensGemma-SEA-LION-v4.5-E2B-IT official model card |
| License declared by the model card | MITGemma-SEA-LION-v4.5-E2B-IT official model card |
Related models
- Gemma-SEA-LION-v4-4B-VL
AI Singapore · SEA-LION v4
A 4B Gemma 3 vision-language model adapted for regional image understanding, visual parsing and tool calls.
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