- Context (input / output)
- 1M / 131K
- Parameters
- 27.8B
- Release date
- 2026-08
- License
- Apache 2.0
- Input modalities
- Text · Image · Video
- Output modalities
- Text
- Architecture
- Gated DeltaNet / Gated Attention + Vision Encoder
- Modes
- Supports thinking and non-thinking modes
- Type
- Dense Vision-Language Model
- Precision
- BF16
- Total parameters
- 27B
- Exact release date
- 2026-08-12
- Weight status
- Open weights released (BF16)
- Default reasoning effort
- xhigh (supports low / medium / xhigh)