--- format: "aidr-story-markdown/v1" id: "0f839cb2f222045c9647a0b448e2457859435347d567e87f40f131210e506f86" canonical_url: "https://aidr.today/0f839cb2?lang=en" title: "LlamaIndex Launches OCRia-1 to Hit New Record in Document Parsing" lang: "en" requested_lang: "en" available_langs: ["en","vi"] translation_fallback: null fallback_fields: [] published_at: "2026-10-22T15:54:29.000Z" category: "Tools" topics: ["llm","devtools"] source_urls: ["https://marketbrief.now/ai/llamaindex-launches-ocria-1-to-hit-new-record-in-document-parsing-6225a307","https://huggingnews.com/ai/llamaindex-launches-ocria-1-to-hit-new-record-in-document-parsing-6225a307"] summary: "LlamaIndex has launched OCRia-1, a 4B parameter vision-language model that achieves a record 87.4 score on the ParseBench benchmark for document parsing. The model processes documents at over 2,800 tokens per second and is now powering all modes of LlamaParse, handling complex enterprise files such as scanned PDFs, unlabeled charts, and nested tables. It also led third-party benchmarks with scores of 92.9 on olmOCR-bench for tables and 84.1 on PulseBench-Tab." --- # LlamaIndex Launches OCRia\-1 to Hit New Record in Document Parsing > [Open the canonical story]() **Published:** 2026-10-22T15:54:29.000Z **Category:** Tools **Topics:** llm, devtools ## Summary LlamaIndex has launched OCRia\-1, a 4B parameter vision\-language model that achieves a record 87\.4 score on the ParseBench benchmark for document parsing\. The model processes documents at over 2,800 tokens per second and is now powering all modes of LlamaParse, handling complex enterprise files such as scanned PDFs, unlabeled charts, and nested tables\. It also led third\-party benchmarks with scores of 92\.9 on olmOCR\-bench for tables and 84\.1 on PulseBench\-Tab\. ## Sources - [Story source]() - [Story source]()