Machine Learning Computer Vision5 min reading time

Jina AI Releases jina-ocr-v1: A 3.4B MoE Document Parser With Built-In Speculative Decoding for Low-Budget GPUs

MarkTechPostPublished Sep 18, 2026

Jina AI has launched jina-ocr-v1, a 3.4B parameter visual document parser optimized for low-budget GPUs like NVIDIA L4. It converts PDFs and scans into Markdown using a novel speculative decoding method for faster, lossless output.

#document-parsing #jina-ai #nvidia-l4 #optical-character-recognition

Jina AI Releases jina-ocr-v1: A 3.4B MoE Document Parser With Built-In Speculative Decoding for Low-Budget GPUs

More in Machine Learning

Google ML Drift Accelerates On-Device AI Inference Across Diverse GPUs

Researchers from Stanford and UC Berkeley Introduce TokenGate: Token-Gated Attention Makes Long-Context Prefill Up to 11.3x Faster

Jack Dorsey’s Block made an AI model that talks to you like an awkward new friend