# OP-GRPO: Efficient Off-Policy GRPO for Flow-Matching Models

arXiv: https://arxiv.org/abs/2604.04142 · HTML: https://mlsanity.com/paper/2604.04142/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- Placed in Diffusion and flow policies for robotics, one of 14 directions in Diffusion models.
- Builds on 39 of its 44 references that are in the map.
- First posted on arXiv 5 Apr 2026.

## Placed in

- Diffusion models / [Diffusion and flow policies for robotics](https://mlsanity.com/diffusion/diffusion-and-flow-policies-for-robotics/)

## Builds on (most cited first, 30 of 39)

- [Denoising Diffusion Probabilistic Models](https://mlsanity.com/paper/2006.11239/) (2020)
- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [Score-Based Generative Modeling through Stochastic Differential Equations](https://mlsanity.com/paper/2011.13456/) (2020)
- [Deep Unsupervised Learning using Nonequilibrium Thermodynamics](https://mlsanity.com/paper/1503.03585/) (2015)
- [Flow Matching for Generative Modeling](https://mlsanity.com/paper/2210.02747/) (2022)
- [Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor](https://mlsanity.com/paper/1801.01290/) (2018)
- [Scaling Rectified Flow Transformers for High-Resolution Image Synthesis](https://mlsanity.com/paper/2403.03206/) (2024)
- [Neural Ordinary Differential Equations](https://mlsanity.com/paper/1806.07366/) (2018)
- [Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flow](https://mlsanity.com/paper/2209.03003/) (2022)
- [Playing Atari with Deep Reinforcement Learning](https://mlsanity.com/paper/1312.5602/) (2013)
- [Wan: Open and Advanced Large-Scale Video Generative Models](https://mlsanity.com/paper/2503.20314/) (2025)
- [HybridFlow: A Flexible and Efficient RLHF Framework](https://mlsanity.com/paper/2409.19256/) (2024)
- [FLUX.1 Kontext: Flow Matching for In-Context Image Generation and Editing in Latent Space](https://mlsanity.com/paper/2506.15742/) (2025)
- [Training Diffusion Models with Reinforcement Learning](https://mlsanity.com/paper/2305.13301/) (2023)
- [Pick-a-Pic: An Open Dataset of User Preferences for Text-to-Image Generation](https://mlsanity.com/paper/2305.01569/) (2023)
- [GenEval: An Object-Focused Framework for Evaluating Text-to-Image Alignment](https://mlsanity.com/paper/2310.11513/) (2023)
- [Human Preference Score v2: A Solid Benchmark for Evaluating Human Preferences of Text-to-Image Synthesis](https://mlsanity.com/paper/2306.09341/) (2023)
- [Diffusion Model Alignment Using Direct Preference Optimization](https://mlsanity.com/paper/2311.12908/) (2023)
- [Flow-GRPO: Training Flow Matching Models via Online RL](https://mlsanity.com/paper/2505.05470/) (2025)
- [PyTorch FSDP: Experiences on Scaling Fully Sharded Data Parallel](https://mlsanity.com/paper/2304.11277/) (2023)
- [DanceGRPO: Unleashing GRPO on Visual Generation](https://mlsanity.com/paper/2505.07818/) (2025)
- [T2I-CompBench++: An Enhanced and Comprehensive Benchmark for Compositional Text-to-image Generation](https://mlsanity.com/paper/2307.06350/) (2023)
- [Cal-QL: Calibrated Offline RL Pre-Training for Efficient Online Fine-Tuning](https://mlsanity.com/paper/2303.05479/) (2023)
- [TextDiffuser: Diffusion Models as Text Painters](https://mlsanity.com/paper/2305.10855/) (2023)
- [Unified Reward Model for Multimodal Understanding and Generation](https://mlsanity.com/paper/2503.05236/) (2025)
- [MixGRPO: Unlocking Flow-based GRPO Efficiency with Mixed ODE-SDE](https://mlsanity.com/paper/2507.21802/) (2025)
- [Open-Sora 2.0: Training a Commercial-Level Video Generation Model in 200k](https://mlsanity.com/paper/2503.09642/) (2025)
- [Hybrid RL: Using Both Offline and Online Data Can Make RL Efficient](https://mlsanity.com/paper/2210.06718/) (2022)
- [Teaching Large Language Models to Regress Accurate Image Quality Scores using Score Distribution](https://mlsanity.com/paper/2501.11561/) (2025)

## Built on by (most cited first, 7 of 7)

- [RAVEN: Real-time Autoregressive Video Extrapolation with Consistency-model GRPO](https://mlsanity.com/paper/2605.15190/) (2026)
- [MARBLE: Multi-Aspect Reward Balance for Diffusion RL](https://mlsanity.com/paper/2605.06507/) (2026)
- [MotionGRPO: Overcoming Low Intra-Group Diversity in GRPO-Based Egocentric Motion Recovery](https://mlsanity.com/paper/2605.05680/) (2026)
- [LeanGRPO: Eliminating Redundant Recomputation in Diffusion RL](https://mlsanity.com/paper/2609.03528/) (2026)
- [λ-Controlled GRPO: Turning Flow-Matching Ratio Instability into a Budgeted Resource](https://mlsanity.com/paper/2609.22041/) (2026)
- [Unified Trajectory Matching Policy Optimization: Diverse T2I Generation and VLA Generalization](https://mlsanity.com/paper/2609.34688/) (2026)
- [VibeEdit: Image Editing with Canvas Instructions](https://mlsanity.com/paper/2610.12229/) (2026)

## Abstract (arXiv)

Post training via GRPO has demonstrated remarkable effectiveness in improving the generation quality of flow-matching models. However, GRPO suffers from inherently low sample efficiency due to its on-policy training paradigm. To address this limitation, we present OP-GRPO, the first Off-Policy GRPO framework tailored for flow-matching models. First, we actively select high-quality trajectories and adaptively incorporate them into a replay buffer for reuse in subsequent training iterations. Second, to mitigate the distribution shift introduced by off-policy samples, we propose a sequence-level importance sampling correction that preserves the integrity of GRPO's clipping mechanism while ensuring stable policy updates. Third, we theoretically and empirically show that late denoising steps yield ill-conditioned off-policy ratios, and mitigate this by truncating trajectories at late steps. Across image and video generation benchmarks, OP-GRPO achieves comparable or superior performance to Flow-GRPO with only 34.2% of the training steps on average, yielding substantial gains in training efficiency while maintaining generation quality.