Search papers, labs, and topics across Lattice.
PaddleOCR-VL-1.6 enhances document parsing by implementing a region-aware data optimization framework that targets under-optimized areas identified in PaddleOCR-VL-1.5. This model not only improves supervision reliability but also incorporates a progressive post-training approach using curated data selection and reinforcement learning. As a result, PaddleOCR-VL-1.6 achieves a state-of-the-art score of 96.33% on OmniDocBench v1.6, outperforming previous benchmarks and competing favorably with leading vision-language models (VLMs).
Targeted optimization in underperforming regions boosts document parsing accuracy to a record 96.33%, setting a new benchmark in the field.
We introduce PaddleOCR-VL-1.6, an upgraded compact document parsing model built upon PaddleOCR-VL-1.5. Although PaddleOCR-VL-1.5 establishes a strong 0.9B baseline, its remaining errors concentrate in under-optimized regions where model behavior is unstable, data coverage is sparse, or supervision is unreliable. Rather than expanding the training corpus indiscriminately, PaddleOCR-VL-1.6 introduces a region-aware data optimization framework that identifies weak regions from the previous model, applies targeted enhancement to these regions, and improves the reliability of supervision signals. It further adopts a progressive post-training recipe based on curated data selection and reinforcement learning, pushing model performance to a higher level through staged optimization. PaddleOCR-VL-1.6 achieves a new state-of-the-art score of 96.33% on OmniDocBench v1.6, demonstrates strong competitiveness against top-tier VLMs, and provides a practical post-training recipe for the PaddleOCR-VL series.
