---
license: mit
library_name: transformers
---
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
MiMo-VL Technical Report
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
## I. Introduction
In this report, we share our efforts to build a compact yet powerful VLM, MiMo-VL-7B. MiMo-VL-7B comprises (1) a native resolution ViT encoder that preserves fine-grained visual details, (2) an MLP projector for efficient cross-modal alignment, and (3) our MiMo-7B language model, specifically optimized for complex reasoning tasks.
The development of MiMo-VL-7B involves two sequential training processes: (1) A four-stage pre-training phase, which includes projector warmup, vision-language alignment, general multi-modal pre-training, and long-context Supervised Fine-Tuning (SFT). This phase yields the MiMo-VL-7B-SFT model. (2) A subsequent post-training phase, where we introduce Mixed On-policy Reinforcement Learning (MORL), a novel framework that seamlessly integrates diverse reward signals spanning perception accuracy, visual grounding precision, logical reasoning capabilities, and human/AI preferences. This phase yields the MiMo-VL-7B-RL model.
We open-source MiMo-VL-7B series, including checkpoints of the SFT and RL model.
We believe this report along with the models will provide valuable insights to develop powerful reasoning VLMs that benefit the larger community.
### 🛤️ During this journey, we find
- **Incorporating high-quality, broad-coverage reasoning data from the pre-training stage is crucial for enhancing model performance.**
- We collect data spanning multimodal perception, knowledge-intensive problems, and tasks demanding strong reasoning capabilities. We identify high-quality queries via quality scoring, employ a large reasoning model to regenerate responses with long CoT, and then apply rejection sampling to secure high-quality responses.
- Distinct from "lite SFT" approaches, we incorporate a substantial volume of this synthetic reasoning data into the later pre-training stages. Notably, the model’s performance on this dataset continued to improve over multiple epochs (e.g., 5), demonstrating resilience against saturation.
- **Mixed On-policy Reinforcement Learning further enhances model performance, while achieving stable simultaneous improvements still remains challenging**
- We apply RL across diverse capabilities, including reasoning, perception, grounding, and human preference alignment, spanning modalities including text, images, and videos. While this hybrid training approach further unlock model’s potential, interference across data domains remains a challenge.
## II. Model Details
> Models are available at [Huggingface Collections: MiMo-VL](https://huggingface.co/collections/XiaomiMiMo/mimo-vl-68382ccacc7c2875500cd212) and [ModelScope Collections: MiMo-VL](https://www.modelscope.cn/collections/MiMo-VL-bb651017e02742)
| **Model** | **Description** | **Download (HuggingFace)** | **Download (ModelScope)** |
| :------------: | :-------------------------------------------------------------------: | :-----------------------------------------------------------------------------: | :---------------------------------------------------------------------------------------: |
| MiMo-VL-7B-SFT | VLM with extraordinary reasoning potential after 4-stage pre-training | [🤗 XiaomiMiMo/MiMo-VL-7B-SFT](https://huggingface.co/XiaomiMiMo/MiMo-VL-7B-SFT) | [🤖️ XiaomiMiMo/MiMo-VL-7B-SFT](https://www.modelscope.cn/models/XiaomiMiMo/MiMo-VL-7B-SFT) |
| MiMo-VL-7B-RL | RL model leapfrogging existing open-source models | [🤗 XiaomiMiMo/MiMo-VL-7B-RL](https://huggingface.co/XiaomiMiMo/MiMo-VL-7B-RL) | [🤖️ XiaomiMiMo/MiMo-VL-7B-RL](https://www.modelscope.cn/models/XiaomiMiMo/MiMo-VL-7B-RL) |
## III. Evaluation Results
### General Capabilities
In general visual-language understanding, MiMo-VL-7B models achieve state-of-the-art open-source results.
### Reasoning Tasks
In multi-modal reasoning, both the SFT and RL models significantly outperform all compared open-source baselines across these benchmarks.
> [!IMPORTANT]
> Results marked with \* are obtained using our evaluation framework.
> Tasks with ${\dagger}$ are evaluated by GPT-4o.
### GUI Grounding
MiMo-VL-7B-RL possess exceptional GUI Understanding and Grounding capabilities. As a general-purpose VL model, MiMo-VL achieves comparable or even superior performance to GUI-specialized models.
### Elo Rating
With our in-house evaluation dataset and GPT-4o judgments, MiMo-VL-7B-RL achieves the highest Elo rating among all evaluated open-source vision-language models, ranking first across models spanning from 7B to 72B parameters.
## IV. Deployment
The MiMo-VL-7B series maintain full compatibility with the `Qwen2_5_VLForConditionalGeneration` architecture for deployment and inference.
## V. Citation
```bibtex
@misc{coreteam2025mimovl,
title={MiMo-VL Technical Report},
author={{Xiaomi LLM-Core Team}},
year={2025},
url={https://github.com/XiaomiMiMo/MiMo-VL},
}
```
## VI. Contact
Please contact us at [mimo@xiaomi.com](mailto:mimo@xiaomi.com) or open an issue if you have any questions.
