# SenseNova-Loop-9B-Preview

> SenseNova-Loop-9B-Preview is a multimodal language model developed for long-context and agentic workflows. It accepts text and images and generates text for applications that combine document understanding, coding, and interaction with external tools.

- **Organisation:** sensenova
- **Category:** Vision-Language Models
- **Parameters:** 9.7B
- **Context length:** 262,144 tokens
- **Licence:** Apache 2.0 (Commercial use OK)
- **Released:** 2026-10-28
- **Hugging Face:** 5.9K downloads (30 days), 321 likes

## Estimated memory

| Quant | ≈ GB |
|---|---|
| Q2_K | 4.6 |
| Q3_K_M | 6.12 |
| Q4_K_M | 7.22 |
| Q5_K_M | 8.22 |
| Q6_K | 9.28 |
| Q8_0 | 11.53 |
| F16 | 20.39 |

## Runs with

- [Transformers](https://github.com/huggingface/transformers)

## About

SenseNova-Loop-9B-Preview is a multimodal language model developed for long-context and agentic workflows. It accepts text and images and generates text for applications that combine document understanding, coding, and interaction with external tools.

The model combines a looped decoder with two-stream manifold-constrained hyper-connections (mHC). A contiguous block of decoder layers is executed twice with shared weights, while learned mappings manage information flow between two residual streams in the looped section. Together, these components provide a deeper execution path without duplicating the repeated block's parameters.

Documented training recipes combine long sequences, multimodal inputs, and supervised task trajectories, with a maximum training sequence length of 262,144 tokens. For the best model performance, we recommend using our documented deployment environment and configuration parameters.

• Shared looped block: 24 decoder layers are applied twice in sequence, producing 56 decoder-layer executions from 32 distinct layers.
• Two-stream mHC: learned mappings route and combine information across two residual streams in the looped section.
• Hybrid attention: the decoder interleaves linear and full attention in a repeating four-layer pattern.
• Long-context multimodal training: documented recipes include text, images, code, and tool trajectories at sequence lengths up to 256K tokens.

## Alternatives

- [Qwen/Qwen3.5-9B](https://selfhost.directory/models/qwen/qwen3.5-9b) (9.7B, Apache 2.0, ≈7.22 GB at Q4): Over recent months, we have intensified our focus on developing foundation models that deliver exceptional utility and performance.
- [Qwen/Qwen3-VL-8B-Instruct](https://selfhost.directory/models/qwen/qwen3-vl-8b-instruct) (8.8B, Apache 2.0, ≈6.82 GB at Q4): Meet Qwen3-VL — the most powerful vision-language model in the Qwen series to date.
- [Zyphra/ZAYA1-VL-8B](https://selfhost.directory/models/zyphra/zaya1-vl-8b) (9.7B, Apache 2.0, ≈6.58 GB at Q4): ZAYA1-VL-8B is a vision-language model (VLM) built upon Zyphra's ZAYA1-8B LLM. It has state-of-the-art performance among VLMs for its size and inference efficiency.
- [Qwen/Qwen3.5-9B-Base](https://selfhost.directory/models/qwen/qwen3.5-9b-base) (9.7B, Apache 2.0, ≈7.22 GB at Q4): Over recent months, we have intensified our focus on developing foundation models that deliver exceptional utility and performance.
- [omlab/VLX-Seek-1.5-10B](https://selfhost.directory/models/omlab/vlx-seek-1.5-10b) (10B, Apache 2.0, ≈7.44 GB at Q4): VLX-Seek-1.5-10B is the open-source 10B model in the VLX-Seek 1.5 family, designed for fine-grained perception and visual grounding in embodied scenarios.
- [XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B](https://selfhost.directory/models/xiaomimimo/mimo-v2.6-distill-qwen-9b) (9.4B, MIT, ≈7.08 GB at Q4): MiMo-V2.6-Distill-Qwen-9B is a 9B agentic model developed by Xiaomi MiMo through supervised fine-tuning of Qwen3.5-9B on MiMo-generated data. It covers coding, general-purpose agent tasks, visual coding, and cybersecurity.

More: https://selfhost.directory/models/sensenova/sensenova-loop-9b-preview/alternatives

## Runs well on

- [GeForce RTX 4060](https://selfhost.directory/models/sensenova/sensenova-loop-9b-preview/on/rtx-4060): Q4_K_M, est. 26 tok/s
- [Apple M4 (16 GB)](https://selfhost.directory/models/sensenova/sensenova-loop-9b-preview/on/apple-m4-16gb): Q6_K, est. 8.3 tok/s
- [Radeon RX 6700 XT](https://selfhost.directory/models/sensenova/sensenova-loop-9b-preview/on/rx-6700-xt): Q8_0, est. 21 tok/s
- [GeForce RTX 3060 12GB](https://selfhost.directory/models/sensenova/sensenova-loop-9b-preview/on/rtx-3060): Q8_0, est. 19 tok/s
- [Arc B580](https://selfhost.directory/models/sensenova/sensenova-loop-9b-preview/on/arc-b580): Q8_0, est. 24 tok/s
- [Tesla T4](https://selfhost.directory/models/sensenova/sensenova-loop-9b-preview/on/tesla-t4): Q8_0, est. 17 tok/s

## Similar size

- [Qwen/Qwen3.5-9B](https://selfhost.directory/models/qwen/qwen3.5-9b) (9.7B)
- [Qwen/Qwen3-VL-8B-Instruct](https://selfhost.directory/models/qwen/qwen3-vl-8b-instruct) (8.8B)
- [Qwen/Qwen2.5-VL-7B-Instruct](https://selfhost.directory/models/qwen/qwen2.5-vl-7b-instruct) (8.3B)
- [google/gemma-3-12b-it](https://selfhost.directory/models/google/gemma-3-12b-it) (12.2B)
- [XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B](https://selfhost.directory/models/xiaomimimo/mimo-v2.6-distill-qwen-9b) (9.4B)
- [meta-llama/Llama-3.2-11B-Vision-Instruct](https://selfhost.directory/models/meta-llama/llama-3.2-11b-vision-instruct) (10.7B)

## FAQ

**How much VRAM does SenseNova-Loop-9B-Preview need?**
About 7.2 GB at Q4_K_M with an 8K context. Q8_0 needs about 12 GB and full precision about 20 GB. Longer contexts need more memory for the KV cache.

**Can SenseNova-Loop-9B-Preview run on the RTX 4060?**
Yes. It fits at Q4_K_M (about 7.0 GB), at an estimated 26 tokens/s. It fits comfortably on the Apple M4 (16 GB), Radeon RX 6700 XT, RTX 3060 12GB.

**Can I use SenseNova-Loop-9B-Preview commercially?**
Yes. It is released under Apache 2.0. Permissive: use, modify and sell freely; keep the notice and licence text.

**What is the context length of SenseNova-Loop-9B-Preview?**
262,144 tokens (256K). The full window needs far more memory than an 8K context.

**How do I run SenseNova-Loop-9B-Preview locally?**
It runs with Transformers. Each has copy-paste examples in its docs.

## Links

- huggingface: https://huggingface.co/sensenova/SenseNova-Loop-9B-Preview
- github: https://github.com/OpenSenseNova/torchtitan

---
Source: https://selfhost.directory/models/sensenova/sensenova-loop-9b-preview