Open-source parser Marker v2 hits 23.7 pages/sec on a single B200, supports 90+ languages
Shruti_0810 · x · 2026-08-24
Datalab shipped Marker v2, an open-source document parsing pipeline that converts PDF, image, DOCX, and PPTX files into clean Markdown. Its strength is scale: up to 23.7 pages/sec on a single B200 GPU, with support for 90+ languages, 100% open source, and built for parsing millions of documents. The repo has 39k stars on GitHub.
Related event: Open-Source Marker v2 Parses 23.7 Pages per Second on a B200 GPU(2 posts)→
More from coding & agent
- AI as a software factory to replace legacy mess and business processes — arieljalali · 2026-08-24
- LLM Agents Invent Tool Arguments: A Sanitization Checklist Before Execution — blaizedsouza · 2026-08-24
- An AI coding agent is an input to an engineering system, not the system itself — ruslansv · 2026-08-24
- Production Guide: Limit Tool Output Size for Agents — blaizedsouza · 2026-08-24
- Six months of evals: never let the model rewrite the source; BM25 weighting lifts recall@10 to 0.86 — Cryvixx · 2026-08-24
- Run AI Tool Parameters in Strict Schema Mode to Catch Prompt Drift — blaizedsouza · 2026-08-24