Show HN: OCR Arena – A playground for OCR models (www.ocrarena.ai)

OCR Arena is a lightweight, interactive playground that pits OCR systems against each other in head-to-head “battles.” Users upload a document (PDF, JPEG, or PNG) to the Battle Arena and the platform runs anonymous models on the same input, producing pairwise comparisons that feed into an Elo-style leaderboard showing ELO, win rate, and number of battles. The interface emphasizes immediacy—drop a file, start an anonymous battle, and watch models compete—while tracking rankings over time. For the AI/ML community this matters because it provides a simple, community-driven way to benchmark OCR performance on real-world inputs. Pairwise battles and Elo scoring make comparisons more dynamic than single-metric leaderboards, helping reveal relative strengths on specific document types, layouts, fonts, or noise conditions. By supporting common file formats and anonymous model entries, OCR Arena can surface edge-case weaknesses, accelerate iterative model tuning, and foster reproducible, crowd-sourced evaluations—useful to researchers, engineers, and product teams looking to compare robustness and practical extraction quality across OCR systems.