One-Step Euler Inference Matches Multi-Step Flow Matching in VLA Models

A CoRL 2026 paper from Araya's RL team shows that flow matching VLA models can achieve multi-step inference quality with just a single Euler integration step, without retraining, speeding up action generation by 1.6 to 4.8 times.

2026-10-26 ~ 2026-10-26 · 2 related posts

• CoRL paper: one-step inference matches multi-step for large VLAs, 1.6-4.8x faster action generation — kaixhin · 2026-10-26
• CoRL paper: flow-matching VLAs can run 1-step Euler inference with no retraining — kaixhin · 2026-10-26