Baidu has released Unlimited-OCR, a model that transcribes dozens of pages in a single forward pass using Reference Sliding Window Attention (R-SWA). It builds on DeepSeek-OCR, inheriting its encoder, image compression, and MoE architecture, with only 500M active parameters per token. The model achieves 93.92% accuracy on OmniDocBench v1.6, outperforming DeepSeek-OCR's 87.01% on v1.5, though vendor-reported results warrant independent validation.
Baidu's Unlimited-OCR Transcribes Dozens of Pages in One Forward Pass
from English