SnapFlow: One-Step Action Generation for Flow-Matching VLAs via Progressive Self-Distillation

Wuyang Luan; Junhui Li; Weiguang Zhao; Rui Ma; Tieru Wu

cs.CVApr 7, 2026

SnapFlow: One-Step Action Generation for Flow-Matching VLAs via Progressive Self-Distillation

Authors: Wuyang Luan, Junhui Li, Weiguang Zhao, Rui Ma, Tieru Wu

Organizations: Jilin University · Chongqing University · University of Liverpool · GenY

Abstract

Vision-language-action (VLA) models with flow-matching action heads, such as π0 and π0.5, have become a standard recipe for generalist robot manipulation, but every action chunk is produced by integrating a learned velocity field over about ten sequential action-expert evaluations, so action generation dominates model latency and caps the achievable control frequency. Existing remedies either shrink each evaluation through layer distillation or pruning while leaving the sequential solver in place, or import one-step objectives from image generation that were designed for training from scratch, rely on an exponential-moving-average teacher or an external critic, and regress finite-span targets onto the conditional velocity, whose variance around the marginal field biases the learned map. We introduce SnapFlow, a self-distillation method that adapts a pretrained flow-matching VLA to one-step action generation. SnapFlow mixes local flow-matching supervision with shortcut samples whose target is the model's own two-step Euler endpoint under stop-gradient, and adds a single zero-initialized destination-time projection so that one action expert serves both objectives while the pretrained predictor is preserved at initialization; no separate teacher network or critic is required. On π0.5, whose 10-NFE teacher operates near ceiling, the archived one-NFE SnapFlow checkpoint reaches 98.75% LIBERO success versus 97.75% for the teacher; under a separate paired evaluation of π0, the one-NFE model surpasses its teacher by 10.5 percentage points. SnapFlow also lowers mean and tail action error on held-out data, cuts batch-one model latency by about 3.3x, and transfers to shelf retrieval on a dual-arm humanoid.

Figures & tables

Figure 1: SnapFlow overview. Top: the destination-time projection augments the pretrained action expert. (a) The teacher uses 10 Euler evaluations. (b) Training mixes local flow-matching supervision with a model-generated two-step endpoint target. (c) Destination $s=0$ produces an action chunk with one expert evaluation.

		LIBERO Success (%)					Latency (A800)
Method	NFE	Spatial	Object	Goal	Long-10	Avg	Model $\downarrow$	Model Speedup
Published references †
Diff. Policy ( Chi et al., 2023 )	100	78.3	92.5	68.3	50.5	72.40	—
Octo-Base ( Ghosh et al., 2024 )	10	78.9	85.7	84.6	51.1	75.08	—
OpenVLA ( Kim et al., 2024 )	AR	84.7	88.4	79.2	53.7	76.50	—
Published $\pi$ 0 ( Black et al., 2024 )	10	97.4	98.4	97.6	93.0	96.60	—

Table 1: LIBERO evaluation under two configurations. The

\pi

0.5 rows reproduce the archived aggregate evaluation; the

\pi

0 rows come from a separate paired evaluation under the configuration used here and are retained as a secondary result. SnapFlow rows use the default training configuration; Section 4.4 evaluates offline training variants. Latency is synchronized batch-one model inference on one A800-80G; speedups are relative to each model’s own 10-NFE teacher. † Reference scores compiled in Appendix E of OpenVLA ( Kim et al., 2024 ) under per-suite fine-tuning, listed for context only.

Method	Avg MSE $\downarrow$	Med MSE $\downarrow$	Std MSE $\downarrow$	P90 MSE $\downarrow$	P95 MSE $\downarrow$	CosSim $\uparrow$
$\pi$ 0.5 — LIBERO (500 samples)
Baseline (10-step)	.01169	.00397	.05412	.01544	.02357	.9885
SnapFlow (1-step)	.00773	.00367	.02964	.01179	.01664	.9916
Relative change: mean $-$ 33.9%, median $-$ 7.6%, std. $-$ 45.2%, P90 $-$ 23.6%, P95 $-$ 29.4%
SmolVLA — PushT
Baseline (10-step)	0.468	0.268	0.517	1.162	—	0.765

Table 2: Extended offline metrics.

\pi

0.5 uses the archived 500-sample LIBERO evaluation; SmolVLA uses 500 PushT samples.

Figure 2: Quality frontier across VLAs and closed-loop success. (a) Offline action MSE against the number of action-expert evaluations, normalized within each model by its 10-NFE reference. Truncating the pretrained

\pi

0.5 solver traces the dashed frontier; one-NFE SnapFlow lands below the 10-NFE reference on the archived

\pi

0.5 and SmolVLA offline sets. (b) LIBERO success for the archived

\pi

0.5 evaluation and the separate

\pi

0 evaluation.

	Configuration			One-NFE quality
Variant	$\alpha$	$\lambda$	Dest.-time conditioning	MSE $\downarrow$	CosSim $\uparrow$	Observation
(a) FM/Consistency Mixing Ratio $\alpha$
Pure consistency	0.0	0.1	✓	.0115	.9876	No FM branch
Consistency-heavy	0.3	0.1	✓	.0088	.9901	70% endpoint samples
Balanced (default)	0.5	0.1	✓	.0077	.9916	Equal branch probability
FM-heavy	0.7	0.1	✓	.0084	.9908	30% endpoint samples

Table 3: Offline ablation study (

\pi

0.5, 500 held-out samples). Each block changes one training choice; otherwise

\alpha\!=\!0.5

\lambda\!=\!0.1

, and destination-time conditioning is enabled.

Figure 3: Real-robot shelf retrieval on a dual-arm AgiBot humanoid. Keyframes at matched fractions of each episode for the 10-NFE $\pi$ 0.5 teacher and for one-NFE SnapFlow on the same shelf configuration. Both policies reach the shelf, grasp the target product, and extract it. The recordings are a deployment demonstration and are not used for success-rate or latency claims.

Appendix figures & tables11 assets

Supplementary material from the paper’s appendix.

Appendix

Symbol	Meaning
$\mathbf{x}_{0}\in\mathbb{R}^{H\times D}$	Action chunk of horizon $H$ and action dimension $D$
$\boldsymbol{\epsilon}\sim\mathcal{N}(\mathbf{0},\mathbf{I})$	Gaussian noise sample
$\mathbf{x}_{t}=(1-t)\mathbf{x}_{0}+t\boldsymbol{\epsilon}$	Point on the linear flow path at source time $t$
$\mathbf{c}$	Context: VLM prefix encoding the observation and instruction
$t\in[0,1]$	Source time; $t\!=\!1$ is noise, $t\!=\!0$ is the clean action
$s\in[0,t]$	Destination time; $s\!=\!t$ gives flow matching, $s\!=\!0$ gives one-step generation

Appendix

Table 4: Notation. Times run from

t\!=\!1

(noise) to

t\!=\!0

(clean action), so integration proceeds backward in

t

and the destination time satisfies

s\leq t

	libero_spatial			libero_object			libero_goal			libero_10 (long)
Task	Base	Naïve	SF	Base	Naïve	SF	Base	Naïve	SF	Base	Naïve	SF
0	100	100	100	100	100	100	100	100	100	90	90	90
1	100	100	100	100	100	100	100	100	100	100	100	100
2	100	100	100	100	100	100	90	90	90	100	100	100
3	100	100	100	100	90	100	80	90	100	100	100	100
4	100	100	100	100	100	100	100	100	100	100	100	100

Appendix

Table 5: Per-task LIBERO success rate (%) for

\pi

0.5. Archived aggregate evaluation with 10 episodes per task; Base is the 10-NFE teacher, SF is one-NFE SnapFlow.

	libero_spatial			libero_object			libero_goal			libero_10 (long)
Task	Base	Naïve	SF	Base	Naïve	SF	Base	Naïve	SF	Base	Naïve	SF
0	80	90	90	70	100	100	90	60	90	50	40	50
1	70	100	100	60	100	80	90	100	100	40	80	70
2	100	100	100	90	100	100	100	90	70	80	80	80
3	90	100	100	90	50	80	80	90	80	70	90	90
4	100	100	100	100	100	100	70	80	100	30	10	40

Appendix

Table 6: Per-task LIBERO success rate (%) for

\pi

0. Separate paired evaluation under the training and evaluation configuration used here; its protocol is distinct from the archived

\pi

0.5 results.

Figure 4: LIBERO success rate by suite. The left panel reports the archived

\pi

0.5 evaluation; the right panel reports the separate

\pi

0 paired evaluation under the configuration used here. Bars show the 10-NFE teacher, naïve one-step inference, and one-NFE SnapFlow.

Method	Layers (architecture)	Steps (sampling)	Success $\Delta$	Model speedup	Mechanism
Shallow- $\pi$ ( Jeon et al., 2026 )	18 $\to$ 6	10	$<1$ pp drop	$>2\times$	Layer distillation
EfficientVLA ( Yang et al., 2025 )	28 $\to$ 22	unchanged	$-$ 0.6 pp	1.93 $\times$	Layer/token pruning + caching
SnapFlow	unchanged	10 $\to$ 1	$+1.0$ pp (archived $\pi$ 0.5)	3.3 $\times$	Destination-time distillation

Appendix

Table 7: Mechanisms of concurrent VLA acceleration methods. Numbers follow the cited papers and are not measured under a common protocol.

Figure 5: Latency decomposition: VLM prefix and action expert. SnapFlow reduces the action expert from 10 sequential evaluations to one, which leaves the fixed VLM prefix as the dominant cost. Batch-one model speedup is about 3.3

\times

for

\pi

0.5 and 3.35

\times

for

\pi

VLA	NFE	Model (ms)	Action Fraction	Model Speedup
$\pi$ 0.5 baseline	1	81.0	26%	3.38 $\times$
	2	103.3	42%	2.65 $\times$
	3	124.4	52%	2.20 $\times$
	5	166.9	64%	1.64 $\times$
	10	274.0	78%	1.00 $\times$
$\pi$ 0.5 SnapFlow	1	83	27%	3.3 $\times$

Appendix

Table 8: Model inference latency vs. NFE (A800, batch size 1). The baseline rows sweep the pretrained solver; the SnapFlow row is the distilled one-NFE model. Speedups are relative to the same model at 10 NFE.

	Teacher (pretrained solver)			SnapFlow (distilled)
NFE	MSE $\downarrow$	P90 MSE $\downarrow$	CosSim $\uparrow$	MSE $\downarrow$	P90 MSE $\downarrow$	CosSim $\uparrow$
1	.00892	.01424	.9911	.00826	.01296	.9911
2	.00918	.01368	.9910	.00723	.01064	.9920
3	.00957	.01342	.9906	.00716	.00989	.9920
5	.01116	.01425	.9888	.00772	.01045	.9917
10	.01238	.01672	.9879	.00864	.01324	.9910

Appendix

Table 9: Offline error versus number of Euler steps (

\pi

0.5, 500 held-out samples, shared noise). Teacher rows truncate the pretrained solver; SnapFlow rows integrate the distilled model with the same step counts, using

s=0

only for the one-NFE row.

Figure 6: Sensitivity to the executed action horizon (archived

\pi

0.5 libero_10 sweep). The two methods trade the lead across horizons, with SnapFlow ahead at 5 executed steps and level with the teacher at 10.

Parameter	Value
SnapFlow
FM/Consistency ratio $\alpha$	0.5
Consistency weight $\lambda$	0.1
Shortcut-target clamp range	$[-20,20]$
Target-time projection	Zero-init linear layer
Training

Appendix

Table 10: SnapFlow training hyperparameters.

Figure 7: Additional real-robot shelf configurations. Keyframes at matched fractions of each episode for the 10-NFE teacher and one-NFE SnapFlow on three further shelf layouts (labelled B, C, and D; layout A appears in Figure 3 ). In every pair both policies reach into the shelf, grasp the target product, and extract it. The recordings are a deployment demonstration and are not used for quantitative claims.

Explore similar work

Fewer Steps, Better Actions: Rethinking Flow-Matching Inference for VLA Policies
Sep 18, 2026Zhipeng Tang, Xinda Chen, Weining Rao +5Efficient VLA Model Inference Vision-Language-Action Models
Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China · University of Chinese Academy of Sciences, Beijing, China · Zhongke Haichuan Intelligent
Toward Real-Time VLAs: Stage-Aware Two-Step Flow Denoising and System-Level Evaluation
Sep 30, 2026Di Wu, Rongtian Shen, Ping Liu +8Robotic Control Flow Matching
Magic-Lab Team, Magiclab Robotics Inc. · Southeast University · Harbin Institute of Technology +3
Mean-Flow based One-Step Vision-Language-Action
Mar 2, 2026Yang Chen, Xiaoguang Ma, Bin ZhaoEfficient VLA Model Inference Vision-Language-Action Models

Sep 18, 2026J/K move · Enter open · S save

Fewer Steps, Better Actions: Rethinking Flow-Matching Inference for VLA Policies

Zhipeng Tang, Xinda Chen, Weining Rao, Xiao Li+4

Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China · University of Chinese Academy of Sciences, Beijing, China · Zhongke Haichuan Intelligent

Efficient VLA Model Inference Vision-Language-Action Models Flow Matching

Open paper AskPDF arXiv

Vision-language-action (VLA) policies based on flow matching generate action chunks through repeated evaluations of an action expert. Increasing the number of integration steps raises inference cost, but does not necessarily improve closed-loop success. We propose Coda, which reallocates part of this integration budget to a single learned endpoint correction. A frozen policy first completes a few-step noise-to-action trajectory; a lightweight Transformer then predicts a demonstration-supervised residual using the candidate action, source noise, and shared observation-prefix cache. Only the corrector is trained. On 50 RoboTwin Easy tasks, five-step Coda improves success from 71.64% to 74.68% over the matched five-step baseline, while reducing forward latency by 30.2% relative to the default ten-step policy. A two-step configuration achieves 71.88% success with a 2.12

\times

speedup. An independent 13-task control shows a 5.69-percentage-point gain at nearly equal latency, supporting correction as an effective alternative to additional integration. The same design also improves frozen official SmolVLA, raising two-step success from 60.8% to 69.4%. These results show that endpoint correction improves the quality-latency trade-off of frozen flow-matching policies.

Abstract

Explore similar work

Fewer Steps, Better Actions: Rethinking Flow-Matching Inference for VLA Policies

Toward Real-Time VLAs: Stage-Aware Two-Step Flow Denoising and System-Level Evaluation

Mean-Flow based One-Step Vision-Language-Action