Translation processing and formats
Yasnora Translate uses local AI, user-owned history, optional terminology and bounded document extraction.
Processing boundary
Translate uses the local wics-general runtime and never selects an external provider. Source text, terminology and the translated result are stored in the signed-in user's tenant-scoped history. AI usage is metered against the user's plan.
Inputs and limits
Direct text is limited to 50,000 characters. Supported uploads are TXT, Markdown, CSV, JSON, searchable PDF and DOCX up to 20 MiB; extracted text has the same 50,000-character limit. Image-only PDFs need OCR before translation.
Terminology, review and export
A glossary can contain up to 50 source-to-target mappings. The model is instructed to preserve names, numbers, links, headings, lists, placeholders and code structure, but the output remains generated text and requires human review. Completed translations can be downloaded as TXT, Markdown or DOCX.