Accurate productions without speed are of limited value.  We need to train and prepare the articulators for rapid productions necessary in order to perform the target sounds automatically in conversation.  To facilitate this we need to adjust how we are modeling.   The diagram below illustrates the progression of our modeling. 
Early models are slower and more numerous.  Models get progressively faster and less frequent.  Our ultimate goal is to do more reps in a shorter period of time with a minimal amount of modeling.

When training for speed:

1. Use sets of five.
2. Use no emphasis or start reducing emphasis.
3. Tap out the set of five with your fingers.
4. Model all five productions on the first set only.
5. Model is then given only on the first finger tap.
6. Use no model when child is clear of his/her task.
7. Use positive reinforcement with decreasing frequency.

 Follow this link to my related blogpost for additional information.

Download the 18 Best Tips for Developing Speech Sounds

Get All 18 Tips in One Free Report