Cohere has released Parse 5, a document parsing tool that converts PDFs, slides, and images into structured Markdown at enterprise scale. They are positioning Parse 5 on price-to-performance, rather than raw accuracy, claiming it offers the best price for a score close to the top.
The tool is a 2.3-billion-parameter vision language model built to preserve structure and meaning in enterprise documents, which often mix tables, diagrams, charts, and formatting. Most tools still drop structure or hallucinate content, and even frontier models break on layout-heavy pages, according to Nils Reimers, VP of AI Search at Cohere.
Parse 5 takes a page as an image, runs it through a single vision-language model pass, and returns structured Markdown, collapsing the OCR-plus-model pipeline most tools run as separate steps. It accepts a PDF, PowerPoint, or JPEG page as a base64-encoded image and returns Markdown in reading order, with tables rendered as HTML, image descriptions, and bounding box coordinates for tables and images.
Cohere’s model has an 8,192-token context window and roughly a 4.6-gigabyte footprint. It supports nine languages, including Arabic, English, French, German, Italian, Japanese, Korean, Portuguese, and Spanish, with stable accuracy, and lower-accuracy zero-shot support elsewhere.
Cohere’s own published benchmark comparison, ParseBench, puts Parse 5 behind three larger, general-purpose frontier models on accuracy. GPT-5.5, Opus 4.8, and Gemini 3.5 Flash all score higher than Parse 5 on the three ParseBench dimensions, with scores of 84.4, 84.3, and 81.8, respectively, compared to Parse 5’s score of 79.2.
However, they are claiming the best price for a score close to the top, with the model priced at $1.50 per 1,000 pages through its API. This pricing means that choosing Parse 5 over a large general-purpose model like GPT-5.5 would reduce costs by more than 98 percent for a large financial services firm that processes 750 million documents a year, which could lead to significant savings for companies dealing with code flaws in their documents.
Kevin Petrie, VP of Research at BARC US, said document parsing sits at the center of enterprise AI adoption right now, with 62% adoption rates among organizations polled. Petrie added that only time will tell how Cohere’s cost-performance stacks up against frontier models, but strategically, his view is that Cohere has the right focus.
Stephanie Walter, Practice Leader for AI Stack at HyperFRAME Research, sees Cohere Parse 5 as sitting in a good spot between legacy OCR and using an expensive frontier model on every page. Its potential advantage is delivering structure, spatial provenance, and private deployment at a price suitable for high-volume ingestion, Walter said.
Walter added that the real test is downstream, not on the benchmark, and that enterprises should test parsers against their own most difficult documents and measure downstream retrieval and task accuracy, not how clean the extracted text looks. The key issue is whether the agent can now use the information correctly.
Parse 5 is generally available now through the Cohere API, Model Vault, Microsoft Foundry, and AWS SageMaker, and its potential to make reliable enterprise-scale parsing economical will be closely watched by enterprises looking to adopt AI solutions, particularly those dealing with international documents that require compliance with new visa rules.
It supports nine languages.
