Action tokenization maps continuous action vectors to a sequence of discrete symbols from a fixed vocabulary, so that an autoregressive model, typically a language-model backbone, can predict actions with the same next-token objective it uses for text. RT-2 (Brohan et al., 2023) established the approach for VLAs by expressing robot actions "as text tokens". The simplest scheme bins each action dimension independently: OpenVLA discretizes "each dimension of the robot actions separately into one of 256 bins", with bin edges set from the 1st and 99th percentiles of the training actions.
Per-dimension, per-timestep binning degrades at high control frequency. FAST (Pertsch et al., "FAST: Efficient Action Tokenization for Vision-Language-Action Models", 2025) found that such schemes "typically perform poorly when learning dexterous skills from high-frequency robot data". FAST instead compresses an action chunk with the discrete cosine transform before tokenizing. π0-FAST is π0 trained with this tokenizer. VQ-BeT learns its vocabulary instead, "tokenizing continuous actions with a hierarchical vector quantization module".
Tokenization sets decoding cost. An autoregressive VLA emits tokens one forward step at a time, so latency grows with the number of tokens per chunk. Continuous heads, such as flow matching, diffusion or the L1 regression with parallel decoding used by OpenVLA-OFT, produce the whole chunk in a fixed number of passes.
A common misreading concerns FAST's headline speed-up. The FAST paper reports matching diffusion VLAs "while reducing training time by up to 5x". That figure is about training, not inference speed, and should not be quoted as a latency advantage of π0-FAST over π0.