← Catalog
SenseNova-Loop-9B-Preview
PreviewSenseTimeOpen source
Looped multimodal (text+image in, text out) model for long-context and agentic workflows: a 24-layer block run twice with shared weights (56 layer executions from 32 distinct layers), two-stream manifold-constrained hyper-connections, and hybrid linear/full attention. ~9B parameters (~14.6B effective at inference); trained at up to 262,144-token sequences; MTP supported. Apache-2.0 weights on Hugging Face.
Specifications
- License
- Open source · Apache-2.0
- Weights
- Downloadable
- Architecture
- hybrid
- Parameters
- 9B
- Context window
- — tokens
- Max output
- —
- Knowledge cutoff
- —
- Price (in / out, $/M)
- —
- Modalities
- TextVisionCode
Benchmarks
No benchmark scores recorded yet. Spotted some? Submit a correction.
Vendor-reported figures are claims until independently verified. See methodology.