SenseNova-SI-1.3-8B
SenseTime · 2B and 8B sizes
An open small model for spatial intelligence
SenseTime’s February 2026 open spatial-intelligence model, in 2B and 8B sizes under Apache-2.0: spatial relations and 3D scene understanding are the strengths, at perception 90; documents 65, generation 40 and agents 55 show it is vertical-only. Composite 69.8 makes it one of this edition’s specialists.
Last updated Oct 21, 2026Editorial scoring
Bottom line
Eight-dimension breakdown
editorial mapping · 0–100- SenseNova-SI-1.3-8B69.8
Weights: perception 20 / documents 18 / charts 14 / video 14 / agents 14 / generation 6 / Chinese 6 / ecosystem 8. Definitions and sources on the methodology page.
Public benchmarks
as reported, not converted| Benchmark / metric | Value | Note |
|---|---|---|
| 参数规模 | 8B | Also a 2B tier |
| 许可证 | Apache-2.0 | Commercial use |
| 上下文 | 128K | Official spec |
Public scores come from different versions and prompts and are not interchangeable with our composite; they are one evidence input.
Scenario mapping
five stress tests, one scale| Scenario | Score | Observation |
|---|---|---|
| Spatial relations | 90 · A new capability axis here | |
| Chinese receipt OCR | 65 · Not a document model | |
| Chart data extraction | 60 · Limited | |
| 30-min video QA | 70 · Works on spatial video | |
| Chinese cultural context | 80 · Average general context |
Strengths and limitations
Strengths
- Spatial relations and 3D scene understanding are scarce among open weights.
- 8B, Apache-2.0 and single-GPU inference suit robot vision pre-processing.
- Grade A: weights and evaluation scripts are public.
Limitations
- Documents 65 and charts 60: cannot carry general image-text work.
- Generation 40 and agents 55: no creative or automation capability.
- Chinese 80; niche scenarios need domain fine-tuning.
Price and real cost
List price
—N/A
Open weights; runs on one GPU. For 1,000 images at 1,024px, ~1,100 image tokens each and 300 output tokens, the monthly cost is about ¥0.0 (assuming input at 25% of output price).
Use your volumeAbout SenseNova-SI-1.3-8B
Is SenseNova-SI-1.3-8B open source?
Licence: Open weights (Apache-2.0). Weights are available, but check the exact terms and filing requirements before commercial use.
How much does SenseNova-SI-1.3-8B cost?
Open weights; runs on one GPU. It ranks #52 with a composite of 69.8. Real bills depend on resolution, tiling, caching and volume discounts.
What is SenseNova-SI-1.3-8B best for?
Spatial-relation judgement; Robot vision pre-processing; AR and 3D scene understanding。
Why does SenseNova-SI-1.3-8B score 69.8?
The composite weights eight dimensions: perception 20%, documents 18%, charts 14%, video 14%, agents 14%, generation 6%, Chinese 6%, ecosystem 8%. Its strongest dimension is Visual perception & grounding; its weakest is Generation & editing.
Personal ranking · Not an official score
Every ranking, score and comparison on this page is an independent personal ranking, not an official score, and represents no vendor, institution or regulator. Scores are a subjective mapping of public evidence under our own weights — change the weights and the order changes. Disagree, or spotted a factual error? Contact the site owner: jizhi2024@foxmail.com — we check each report and log corrections in the changelog.