deeplang-ai/LingoEDU-4B
TEXT GENERATIONConcurrency Cost:1Model Size:4BQuant:BF16Ctx Length:32kPublished:Dec 16, 2025License:mitArchitecture:Transformer0.0K Open Weights Warm

LingoEDU-4B by deeplang-ai is a 4 billion parameter model, specialized from Qwen3-4B, designed for advanced document structure analysis. It transforms linear discourse into a condensed hierarchical tree, anchoring nodes to source text via coordinate pointers. With a 40960 token context length, it excels at extracting and structuring information from text documents, outperforming general LLMs and parser APIs on the StructBench dataset in both structural accuracy (TED) and data accuracy (DLA), while maintaining low latency and cost. Its primary use case is precise document parsing and hierarchical information extraction.

Loading preview...