Settings

Model and pipeline parameters. Changes apply immediately to new extractions.

LLM Models

Main LLM Model (extraction, analysis, translation)

Current: gpt-4.1

OCR Model (vision-based OCR for scanned PDFs)

Current: gpt-4.1

Embeddings Model (for RAG vector store)

Current: voyage-4

Temperature (extraction)

Top-K chunks (RAG hits per field)

Max candidates (hypothesize stage)

Snippet chars (context per hit)

Reasoning effort (for o-series and reasoning models)

Current: none

Provider