OCRia-1 (LlamaIndex) — A 4B Parser with the Biggest Jumps on Tables & Charts; Weights Not Yet Opened
OCRia-1 is a 4B VLM (Qwen3.5-4B base) for parsing document pages into Markdown — tables as HTML with full structure, charts as data-point tables, formulas as LaTeX. ParseBench claims: overall score **87.4** (12.1 points above LightOnOCR-3-4B's 75.3); tables dimension 91.5; charts dim…
OCRia-1 is a 4B VLM (Qwen3.5-4B base) for parsing document pages into Markdown — tables as HTML with full structure, charts as data-point tables, formulas as LaTeX.
ParseBench claims: overall score **87.4** (12.1 points above LightOnOCR-3-4B's 75.3); tables dimension 91.5; charts dimension **85.1** (19 points above the runner-up — conventional parsers score near zero on this dimension); formatting 83.4.
It also leads on olmOCR-bench tables (92.9), PulseBench-Tab (84.1), and fr-bench-pdf2md long-table (83.2).
Speed: ~4,700 tokens/sec single-request on a single B200 (claimed 9× faster…