SenseNova U1 Models by SenseTime
2 models2026
Last refreshed 2026-06-29. Next refresh: weekly.
Details
ResearcherSenseTime
LicenseApache 2.0OSI-approved
Commercial useCommercial use: permitted
Models2
Released2026
Capabilities
VisionAll models
MultimodalAll models
About
SenseNova U1 is SenseTime's open-source unified multimodal model family, released April 2026. Built on the NEO-Unify architecture that eliminates both the visual encoder and VAE to process text and images in a single representation space — the first commercially viable model with native unified image understanding and generation. Available under Apache 2.0.
Decision facts
- Best fit
- chinavision and multimodal workcoding
- Capability starting point
- SenseNova-U1-8B with multimodal inputs
- Lowest tracked input
- Not tracked
- Closest related family
- Unisound U2
Current Variants
Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.
2 in view
SenseNova-U1-8BCurrent
Use when the workload needs 8B parameters, multimodal inputs, and china.
2026-048B parametersmultimodal inputschina
SenseNova-U1-A3BCurrent
Use when the workload needs 3B parameters, multimodal inputs, and china.
2026-043B parametersmultimodal inputschina
| Model | Use when | Released | Signals | Status |
|---|---|---|---|---|
| SenseNova-U1-8B | Use when the workload needs 8B parameters, multimodal inputs, and china. | 2026-04 | 8B parametersmultimodal inputschina | Current |
| SenseNova-U1-A3B | Use when the workload needs 3B parameters, multimodal inputs, and china. | 2026-04 | 3B parametersmultimodal inputschina | Current |
Release Timeline
1 release group2026-04
2 current
SenseNova-U1-8B
Current8B parametersmultimodal inputschina
SenseNova-U1-A3B
Current3B parametersmultimodal inputschina
Specifications(2 models)
| Model | Released | Parameters | Vision | Multimodal |
|---|---|---|---|---|
| SenseNova-U1-8B | 2026-04 | 8B | Yes | Yes |
| SenseNova-U1-A3B | 2026-04 | 3B | Yes | Yes |
