FAQ
How does Pulpie achieve 20x cost reduction?
Pulpie uses an encoder architecture that processes HTML in a single forward pass, labeling each block as boilerplate or content. Traditional decoders generate output token-by-token, requiring a full model reload per token. This makes Pulpie compute-bound instead of memory-bound, enabling massive savings at scale.
What formats does Pulpie output?
Clean HTML (preserving structure) or Markdown (for indexing, RAG systems, training data). Specify output format per API call.
Is there a free tier?
Yes. New accounts receive 10M free extracted pages to test quality and speed. After that, pay per consumption.
How accurate is extraction?
Pulpie matches state-of-the-art accuracy (98%+ precision on boilerplate removal). Optimized for production web: news, blogs, documentation, e-commerce.
What's the uptime SLA?
Professional: 99.5%. Enterprise: 99.9% with dedicated infrastructure and 24/7 support.
How fast is integration?
Minutes for REST API (generate key, make request). Batch processing handled within 15 minutes (Professional) or faster (Enterprise SLA).
Do you offer on-premise deployment?
Yes, available on Enterprise plans. Air-gapped and fully private deployments supported.