Paraphrase-multilingual-MiniLM-L12-v2, Table Transformer Detection, and Bielik-11B-v3.0-Instruct are now available in Amazon SageMaker JumpStart

Amazon Web Services has expanded its SageMaker JumpStart model catalog with three new AI models targeting different enterprise use cases. The additions include paraphrase-multilingual-MiniLM-L12-v2, a lightweight semantic similarity model that supports over 50 languages for cross-lingual search and document clustering; Microsoft's Table Transformer Detection model for identifying tables in unstructured documents like PDFs; and Bielik-11B-v3.0-Instruct, an 11-billion parameter language model specialized for European languages with particular strength in Polish. The paraphrase-multilingual-MiniLM-L12-v2 model from Sentence Transformers converts text into 384-dimensional vectors for semantic analysis across multiple languages without requiring language-specific configuration. Microsoft's Table Transformer Detection uses a DETR-based architecture trained on the PubTables-1M dataset to locate tabular content in research papers, financial reports, and other documents. Meanwhile, the Bielik model, developed by SpeakLeash and ACK Cyfronet AGH, covers 32 European languages and is designed for dialogue, mathematical reasoning, and enterprise applications requiring nuanced European language understanding. Customers can deploy these models through SageMaker Studio's interface or the SageMaker Python SDK, allowing organizations to integrate multilingual semantic search, automated document processing, and European language AI capabilities into their existing workflows with minimal setup requirements.