Contextual Flow Matching: Adaptive Step Selection in Flow Models for Efficient Visual Generation

Divya Jyoti Bajpai; Arun Verma; Manjesh Kumar Hanawal

cs.CVOct 2, 2026

Contextual Flow Matching: Adaptive Step Selection in Flow Models for Efficient Visual Generation

Authors: Divya Jyoti Bajpai, Arun Verma, Manjesh Kumar Hanawal

Organizations: Department of IEOR IIT Bombay, Mumbai, India · SMART Centre Singapore

Abstract

Flow Matching enables high-quality visual generation via continuous-time dynamics, but inference remains costly due to multiple sequential function evaluations. Existing acceleration methods reduce the number of function evaluations but often introduce additional training overhead, degrade quality, or fail to account for input-dependent variability. We propose COFLOW, an inference-time method that adaptively selects the step counts each generation based on the prompt features. Our context-aware COFLOW is trained online with an unsupervised reward that balances inference efficiency and generation fidelity. Our method is plug-and-play, requiring no retraining of the underlying generative model. It generalizes to image and video generation, achieving over 2.5x speedup while preserving perceptual and semantic quality. We further provide a theoretical analysis establishing an O(1/K) forward-Euler discretization error bound under standard regularity conditions.

Figures & tables

Figure 1: Generation across integration step budgets. Different inputs can exhibit different behavior as the integration budget changes. The rightmost plots illustrate relative velocity variation along the corresponding trajectories.

Figure 2: Overview of our method. CoFlow uses prompt context to adaptively select the integration step budget, allocating computation according to the learned surrogate reward while retaining the original flow trajectory and solver schedule.

Method	BAGEL				FLUX
	Overall $\uparrow$	IQA $\uparrow$	NFE $\downarrow$	Time $\downarrow$	Overall $\uparrow$	IQA $\uparrow$	NFE $\downarrow$	Time $\downarrow$
Full Model
Full-50	0.78	0.65	50	76.4	0.66	0.75	50	34.1
Full-25	0.75	0.62	25	40.9	0.64	0.71	25	17.3
Full-10	0.71	0.54	10	18.9	0.57	0.63	10	7.26
Static Baselines

Table 1: Comparison of methods on BAGEL and FLUX. We report generation quality (GenEval, IQA) along with efficiency metrics (number of function evaluations and inference time).

Panel of Figure 3 — Figure 3: Comparison of edit quality across two models: BAGEL and FLUX. Each subfigure reports semantic consistency (G_SC), perceptual quality (G_PQ), and overall score (G_O) versus speedup.

Figure 4: Comparison of video generation score and quality where we report the VBench score and the BRISQUE metric that assesses the quality of video.

Appendix figures & tables17 assets

Supplementary material from the paper’s appendix.

Appendix

Figure 5: Comparison of edit quality on Step-1X-Edit model. Each subfigure reports semantic consistency (G_SC), perceptual quality (G_PQ), and overall score (G_O) versus speedup.

Figure 6: L1-relative velocity variation across integration trajectories. Velocity-change profiles vary across prompts and integration budgets, illustrating the trajectory-level signal used by CoFlow f. Quantitative analysis against a 100-step reference is provided in Appendix A.6 .

Method	Time (s)
Full Model
Full-50	75.6
Full-25	40.1
Ours
NCB-50	76.4
NCB-25	40.9

Appendix

Table 2: Inference time comparison of CoFlow .

Figure 7: Qualitative examples showcasing the choices of AdaDiff and CoFlow .

Method	Compositional Dimensions						Overall Metrics
	SO	TO	CO	CL	CA	PO	Overall	CLIPIQA	NFE
Full Model
Full-50	0.99	0.90	0.81	0.85	0.59	0.54	0.78	0.65	50
Full-25	0.99	0.90	0.76	0.81	0.60	0.49	0.75	0.62	25
Full-10	0.99	0.85	0.67	0.78	0.56	0.46	0.71	0.54	10
Static Baselines

Appendix

Table 3: Comparison across various Compositional Dimensions and overall performance for the BAGEL model. Our method achieves strong performance while significantly reducing the number of function evaluations (NFE).

Method	Compositional Dimensions						Overall Metrics
	SO	TO	CO	CL	CA	PO	Overall	CLIPIQA	NFE
Full Model
Full-50	0.97	0.79	0.75	0.78	0.48	0.20	0.66	0.75	34.1
Full-25	0.97	0.78	0.71	0.75	0.45	0.18	0.64	0.71	17.3
Full-10	0.97	0.65	0.58	0.67	0.41	0.15	0.57	0.63	7.26
TeaCache	0.97	0.76	0.73	0.74	0.44	0.18	0.63	0.72	20.8

Appendix

Table 4: Comparison of the Compositional Dimensions for the FLUX model. Our method maintains competitive compositional performance while reducing computation (NFE).

Analysis	Spearman $\rho$
Pooled prompt–budget pairs	0.822
Mean within-prompt	0.924
Median within-prompt	0.970
Mean cross-prompt at fixed budget	0.255

Appendix

Table 5: Association between L1-relative velocity variation and deviation from a 100-step reference. All step budgets for each prompt use the identical initial latent.

Method	Overall $\uparrow$	IQA $\uparrow$	NFE $\downarrow$	Time $\downarrow$
Ours + UCB (No Context)	0.74	0.60	20.2	31.9
Ours + UCB	0.77	0.65	17.7	27.5
Ours + TS	0.76	0.65	17.5	26.9

Appendix

Table 6: Ablation of context and bandit strategy. Conditioning the policy on input context improves the quality–efficiency trade-off relative to the non-contextual UCB variant, while UCB and Thompson Sampling provide comparable contextual policies.

Context	Mean NFE $\downarrow$	Max-step rate $\downarrow$
No context	39.75	65.4%
Handcrafted only	37.79	52.9%
MiniLM only	39.58	60.0%
Full	37.67	47.9%

Appendix

Table 7: Effect of context representation. Results are aggregated over three seeds. Max-step rate denotes the fraction of decisions selecting the largest available step budget.

Rounds	Reward $\uparrow$	NFE $\downarrow$	Max-step rate
1–10	0.520	43.2	72.0%
11–20	0.548	40.5	63.3%
21–40	0.566	38.7	55.8%
41–60	0.575	37.9	51.6%
61–80	0.578	37.6	50.4%
81–100	0.580	37.4	49.8%

Appendix

Table 8: Online adaptation and order robustness. Results are aggregated over five random prompt permutations and three independent seeds per permutation. Max-step rate denotes the fraction of decisions selecting the largest available step budget.

Policy	Reward $\uparrow$	NFE $\downarrow$	Max-step rate	Action entropy $\uparrow$
Greedy	$0.541\pm 0.011$	$39.6\pm 1.2$	$63.4\%$	$1.21$
NeuralUCB	$\mathbf{0.582\pm 0.008}$	$\mathbf{37.5\pm 0.8}$	$49.7\%$	$\mathbf{1.54}$

Appendix

Table 9: Effect of UCB exploration. NeuralUCB is compared with an otherwise identical online greedy policy with the uncertainty bonus removed after initialization. We report mean

\pm

standard deviation over three independent seeds. Action entropy denotes the entropy of the empirical distribution of selected step budgets.

Method	Compositional Dimensions						Overall Metrics
	SO	TO	CO	CL	CA	PO	Overall	CLIPIQA	NFE
Full Model
Full-10	0.99	0.79	0.37	0.88	0.21	0.15	0.57	0.72	10
Full-6	0.99	0.78	0.36	0.86	0.20	0.15	0.55	0.68	6
Full-3	0.98	0.75	0.29	0.85	0.18	0.12	0.52	0.62	3
Ours

Appendix

Table 10: Comparison under low-step regimes where the PeRFlow model, which is trained for low step generation within

5

8

10

steps, where our method further improves this method by making the choice of number of steps adaptive. Our method achieves competitive performance while reducing the number of function evaluations (NFE).

Figure 8: Direct step reduction vs. trajectory truncation. Direct reduction generates with a timestep schedule constructed for the reduced step budget, whereas trajectory truncation obtains a comparable computational reduction by skipping intermediate evaluations from a higher-step trajectory.

Method	GenEval $\uparrow$	CLIP-IQA $\uparrow$	NFE $\downarrow$	Time $\downarrow$
Baseline Methods
TeaCache	0.74	0.59	29.4	45.8
FastFlow	0.75	0.59	20.9	32.9
FlowCast	0.76	0.60	21.5	34.2
With Ours (Adaptive Step Selection)
TeaCache + Ours	0.71	0.55	16.8	25.7

Appendix

Table 11: Combination with existing acceleration methods on BAGEL. Adaptive step-budget selection can be applied together with TeaCache, FastFlow, and FlowCast, providing additional runtime reductions with method-dependent changes in generation quality.

Hyperparameter	FLUX		BAGEL
Hyperparameter	Generation	Editing	Generation	Editing
Image resolution	$1360\times 768$	–	$1024\times 1024$	–
Guidance scale	$2.5$	$2.5$	$4.0$	$4.0$
Guidance rescale	$0$	$0$	$1.0$	$2.0$
$\mu$	$\tfrac{1}{50}$	$\tfrac{1}{60}$	$\tfrac{1}{50}$	$\tfrac{1}{60}$
CFG interval	–	–	$[0.4,\,1.0]$	$[0.0,\,1.0]$

Appendix

Table 12: Hyperparameter settings for FLUX and BAGEL across generation and editing tasks.

Method	Characteristics	GenEval	IQA	NFE
Base Solvers (50 Steps)
Euler	Deterministic, 1st-order	0.60	0.71	50
Euler-anc	Stochastic, 1st-order	0.57	0.72	50
DPM++2M	Deterministic, multistep	0.59	0.70	50
DPM++2S-anc	Stochastic, multistep	0.57	0.71	50
Heun	Deterministic, 2nd-order	0.59	0.70	50

Appendix

Table 13: Comparison across solver families on a Stable Diffusion backbone. We categorize solvers by stochasticity and numerical integration type. CoFlow consistently reduces the number of function evaluations (NFE) by

\sim

2.5

\times

while maintaining comparable generation quality.

Feature	Definition	Description
Word count	$\min\left(\frac{\#\text{words}}{50},\,1\right)$	Captures prompt length and compositional richness. Longer prompts typically involve more entities, relations, or attributes, increasing reasoning complexity.
Subword diversity (TTR)	$\frac{\|\text{unique tokens}\|}{\#\text{tokens}}$	Measures lexical diversity. Higher diversity indicates richer semantics and reduced redundancy, often correlating with more complex generation requirements.
Mean word length	$\min\left(\frac{1}{10}\cdot\frac{\sum\|w\|}{\#\text{words}},\,1\right)$	Acts as a proxy for linguistic specificity. Longer words often correspond to more descriptive or specialized concepts.
Numeric fraction	$\frac{\#\text{words containing digits}}{\#\text{words}}$	Captures numerical constraints (e.g., counts or quantities), which increase compositional difficulty and require precise control in generation.
Object density (POS-based)	$\min\left(\frac{\#\text{nouns}}{\#\text{words}},\,1\right)$	Measures the number of entities in the prompt. Higher density implies more objects and interactions, increasing spatial and compositional complexity.
Attribute density (POS-based)	$\min\left(\frac{\#\text{adjectives}}{\#\text{words}},\,1\right)$	Captures descriptive richness. More attributes introduce finer-grained constraints (e.g., color, size), making generation more demanding.

Appendix

Table 14: Handcrafted prompt features used alongside MiniLM embeddings to construct the context vector

\varphi(x)

. The features provide lightweight compositional, lexical, numerical, and structural cues for contextual step selection.

Explore similar work

Perceptual Flow Matching for Few-Step Generative Modeling
Jul 3, 2026Chuyang Zhao, Yifei Song, Hongfa Wang +7Flow Matching Perceptual Loss
1Joy Future Academy · 2Fudan University · 3Tsinghua University +1
SubFlow: Sub-mode Conditioned Flow Matching for Diverse One-Step Generation
Apr 14, 2026Yexiong Lin, Jia Shi, Shanshan Ye +3Flow Matching Flow-Based Generative Modeling
Sydney AI Centre, The University of Sydney, Sydney, NSW, Australia · School of Artificial Intelligence, Xidian University, Xi’an, Shaanxi, China · Australian Artificial Intelligence Institute, University of Technology Sydney, Sydney, NSW, Australia +1
CAT-Flow: Curvature-Adaptive sTeps for Flow Matching
Sep 1, 2026Qinchan Li, Pedro Cisneros-Velarde, Keru Fu +3Flow Matching T2I Generation
Simon Fraser University · VMware Research

Jul 3, 2026J/K move · Enter open · S save

Perceptual Flow Matching for Few-Step Generative Modeling

Chuyang Zhao, Yifei Song, Hongfa Wang, Jianlong Yuan+6

1Joy Future Academy · 2Fudan University · 3Tsinghua University+1

Flow Matching Perceptual Loss Image Generation Video Generation Flow-Based Generative Modeling

Open paper AskPDF arXiv

We propose Perceptual Flow Matching (PFM), a simple yet effective framework for few-step generation in flow-matching models. Rather than performing velocity regression in the conventional VAE latent space, PFM supervises flow matching in a perceptual feature space using pretrained perceptual models. This simple change substantially improves the few-step generation capability of flow-matching models, reducing the number of sampling steps from 35-50 to 4-8 while preserving generation quality. Unlike existing acceleration and distillation approaches, PFM requires neither teacher models nor auxiliary score networks and can be integrated into standard flow-matching training pipelines with minimal modifications. Extensive experiments on image generation, video generation, and image editing tasks demonstrate that PFM consistently produces high-quality results while producing fewer artifacts than existing distillation-based methods. We further show that perceptual supervision shifts the regression minimizer from mean-seeking to mode-seeking, biasing predictions toward on-manifold modes that remain accurate under coarse few-step integration. Our results reveal that standard flow-matching training can naturally yield high-quality few-step generators when supervised in an appropriate representation space. We hope this insight inspires future research into representation-aware objectives for efficient generative modeling.

Abstract

Explore similar work

Perceptual Flow Matching for Few-Step Generative Modeling

SubFlow: Sub-mode Conditioned Flow Matching for Diverse One-Step Generation

CAT-Flow: Curvature-Adaptive sTeps for Flow Matching