# UniEvo-VL: An On-policy Self-Distillation Training Recipe for Multimodal Model Self-improvement

arXiv: https://arxiv.org/abs/2609.38721 · HTML: https://mlsanity.com/paper/2609.38721/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Self-improving MLLMs and synthetic data, one of 13 directions in Synthetic data and self-improvement for LLMs.
- Builds on 22 of its 26 references that are in the map.
- First posted on arXiv 30 Sep 2026.

## Placed in

- Synthetic data and self-improvement for LLMs / [Self-improving MLLMs and synthetic data](https://mlsanity.com/synth/self-improving-mllms-and-synthetic-data/)
- Knowledge distillation / [Multimodal distillation](https://mlsanity.com/distill/multimodal-distillation/)

## Builds on (most cited first, 22 of 22)

- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [LoRA: Low-Rank Adaptation of Large Language Models](https://mlsanity.com/paper/2106.09685/) (2021)
- [Qwen3-VL Technical Report](https://mlsanity.com/paper/2511.21631/) (2025)
- [Qwen-Image Technical Report](https://mlsanity.com/paper/2508.02324/) (2025)
- [GenEval: An Object-Focused Framework for Evaluating Text-to-Image Alignment](https://mlsanity.com/paper/2310.11513/) (2023)
- [On-Policy Distillation of Language Models: Learning from Self-Generated Mistakes](https://mlsanity.com/paper/2306.13649/) (2023)
- [Flow-GRPO: Training Flow Matching Models via Online RL](https://mlsanity.com/paper/2505.05470/) (2025)
- [Self-Distilled Reasoner: On-Policy Self-Distillation for Large Language Models](https://mlsanity.com/paper/2601.18734/) (2026)
- [Reinforcement Learning via Self-Distillation](https://mlsanity.com/paper/2601.20802/) (2026)
- [DiffusionNFT: Online Diffusion Reinforcement with Forward Process](https://mlsanity.com/paper/2509.16117/) (2025)
- [From Reflection to Perfection: Scaling Inference-Time Optimization for Text-to-Image Diffusion Models via Reflection Tuning](https://mlsanity.com/paper/2504.16080/) (2025)
- [GenEval 2: Addressing Benchmark Drift in Text-to-Image Evaluation](https://mlsanity.com/paper/2512.16853/) (2025)
- [DiffusionOPD: A Unified Perspective of On-Policy Distillation in Diffusion Models](https://mlsanity.com/paper/2605.15055/) (2026)
- [Flow-OPD: On-Policy Distillation for Flow Matching Models](https://mlsanity.com/paper/2605.08063/) (2026)
- [UniCorn: Towards Self-Improving Unified Multimodal Models through Self-Generated Supervision](https://mlsanity.com/paper/2601.03193/) (2026)
- [SRUM: Fine-Grained Self-Rewarding for Unified Multimodal Models](https://mlsanity.com/paper/2510.12784/) (2025)
- [D-OPSD: On-Policy Self-Distillation for Continuously Tuning Step-Distilled Diffusion Models](https://mlsanity.com/paper/2605.05204/) (2026)
- [UniReason 1.0: A Unified Reasoning Framework for World Knowledge Aligned Image Generation and Editing](https://mlsanity.com/paper/2602.02437/) (2026)
- [Turning Internal Gap into Self-Improvement: Promoting the Generation-Understanding Unification in MLLMs](https://mlsanity.com/paper/2507.16663/) (2025)
- [On-Policy Self-Distillation in Diffusion Models](https://mlsanity.com/paper/2608.24646/) (2026)
- [Meta-TTRL: A Metacognitive Framework for Self-Improving Test-Time Reinforcement Learning in Unified Multimodal Models](https://mlsanity.com/paper/2603.15724/) (2026)
- [Dreaming in Flow: Generative Grounding Feedback for Self-Evolving Unified Multimodal Models](https://mlsanity.com/paper/2609.08282/) (2026)

## Abstract (arXiv)

Modern multimodal models bring generation and understanding into a single unified system, which enables them to provide and learn from their own feedback. Motivated by this unified capacity, we introduce UniEvo-VL, a self-evolving framework for multimodal models to learn from this constructive self-correction feedback during test-time compute. Instead of relying on a separate, often larger, teacher, we leverage their self-critiques as privileged information and ask a single multimodal model to act as both teacher and student with different contexts. The student only sees the vanilla question, while the teacher conditions on the privileged critique. Then training minimizes the per-state divergence between their denoising diffusion distributions over the student's own sampling trajectories. Experiments demonstrate that UniEvo-VL improves the image generation capabilities of multimodal models, while maintaining their sensitivity to additional reflection information. Specifically, we build on top of the open-source Qwen-image-2512 and observe a significant performance gain from 0.747 to 0.808 on GenEval and from 32.97 to 35.53 on GenEval2 Soft-TIFA. Moreover, attempts with more powerful external critics (e.g., GPT5.6-Luna) show that multimodal models with strong judge capabilities can anticipate a higher self-evolving ceiling. Last but not least, mixed text-rendering outcomes show that our self-improvements may not be uniform across different tasks. Our study aims to shed light on the current hot recursive self-improvement research line to enhance the user experience when using multimodal models without external supervision or guidance.