OXR tops all 3 doc parsing benchmarks, 100 pages in 101 seconds at 1.3B params
JaynitMakwana · x · 2026-10-23
OXR recognizes 13 document element types including text, tables, formulas, and chemical structures. Recognition and reading order run in parallel inside the model, while layout analysis stays a swappable external step. It ranks #1 on all three benchmarks: olmOCR-Bench 87.84, OmniDocBench 97.18, DocParsingBench 87.76. The closest model on olmOCR-Bench, Infinity-Parser2-Pro (87.60), takes over 6,400 seconds for 100 pages; OXR does it in 101 seconds with just 1.3B parameters.
Related event: Open-Source 1.3B Document Parsing Model OXR Tops Three Benchmarks(6 posts)→
More from Models
- GPT-4 once gave top-5 token probs with no hidden CoT — researcher says it could've been cloned for under $10M — carrigmat · 2026-10-23
- Google Ultra users report Argon eating 20-30% of usage limit per prompt — Error_33_6070 · 2026-10-23
- Voice bot team: GPT Live 1 and Gemini 3.8 Thinking both fail in production — Mr_Nice_ · 2026-10-23
- KURE rerankers demoed: 149M nano model boosts the right answer to top spot — tomaarsen · 2026-10-23
- KURE-v3 debuts: first model suite covering all 4 Sentence Transformers types — tomaarsen · 2026-10-23
- Debate: are abliterated models causing more critical vulns, or just democratizing their discovery? — max_paperclips · 2026-10-23