LlamaIndex releases OCRia-1, a 4B document parsing model scoring 87.4 on ParseBench
llama_index · x · 2026-10-25
LlamaIndex introduced OCRia-1, a 4-billion-parameter VLM for document parsing that it claims is SOTA on accuracy, cost, and latency, already powering LlamaParse.
- Scores 87.4 on ParseBench, the best for any parsing model under 35B parameters, 12.1 points ahead of LightOnOCR-3-4B (75.3)
- Leads third-party table benchmarks: olmOCR-bench (83.5 overall, 92.9 on tables), PulseBench-Tab (84.1), fr-bench-pdf2md (72.1)
- Excels at complex tables (nested headers, merged cells, scanned grids) and unlabeled charts, while preserving formatting
It outperforms Mistral OCR 4 (67.5), Chandra-OCR-2 (74.8), and olmOCR-2 7B (43.3).
Related event: LlamaIndex Releases OCRia-1, a 4B Document Parsing Model Claiming SOTA(2 posts)→
More from Models
- Claude finds critical vulnerabilities in closed-source binaries as if reading source code — tianyin_xu · 2026-10-25
- A downloadable open model now outperforms one once deemed too dangerous to release — jeetwanderer · 2026-10-25
- Open-source decision model Laya hits 32K phones in one month at ~33ms per call — HowDevelop · 2026-10-25
- New paper: AI writing is optimized for information density, not human communication — steverathje2 · 2026-10-25
- User reports dot's web fetch stays restricted after removing a custom connector — MotherIsland8888 · 2026-10-25
- $0.48 and 43 minutes: DeepSeek V4.1 builds an interactive map, igniting cost vs capability debate — teortaxesTex · 2026-10-25