Cohere Parse, a vision language model designed for enterprise document data conversion, is now accessible on Microsoft Foundry. This model joins Cohere's existing Embed and Rerank models within Foundry. Parse excels at transforming complex, multimodal documents into structured, machine-readable Markdown. It recognizes and interprets tables, forms, diagrams, and images, extracting semantic meaning beyond simple text. The model preserves document structure by providing bounding boxes for visual elements, aiding in retrieval and citation. Cohere specifically trained Parse on a diverse range of enterprise documents across sectors like finance and insurance. It also supports nine major languages, making it globally applicable. Designed for efficiency, Parse can process approximately 4.5 pages per second per GPU, enabling high-throughput workloads. Pricing for Cohere Parse v5 is set at $1.50 per 1,000 pages. Its deployment within Foundry provides enhanced security and governance by keeping inference within a user's Azure tenant boundary.