14x

Read news on 14x with our app.

Read more in the app

Consistency diffusion language models: Up to 14x faster, no quality loss