Open Models Almanac
Methodsinference & servingserving parallelism

single mention · inference & serving

Disaggregated serving across encoding, prefill, and decode

Separates vision encoding, prefill, and decoding so they can scale independently and overlap in execution.

The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See serving parallelism for the methods filed beside it.

Also called encoder-prefill-decode disaggregation.

Filed under inference & serving › serving parallelism