I had a question regarding credit usage for SAM-based annotation (SAM2/SAM3) in Roboflow.
The documentation mentions that roughly 1 credit corresponds to ~100 images. However, I’m trying to better understand how this applies to interactive workflows:
- Does “100 images per credit” mean ~100 model calls?
- For SAM usage, is each click/prompt interaction counted as a separate call toward credits?
- If we refine masks with multiple clicks on the same image, does that consume additional credits?
- How is credit usage handled specifically under the Core plan?
I’m trying to estimate usage more accurately for an annotation workflow, so any clarification would be really helpful.
Thanks in advance!