Cohere model releases

Cohere's enterprise document-intelligence vision-language model, released Aug 27 2026 (parse-v5.0). A compact 2.3B-parameter VLM that converts PDFs, slides, and images into structured Markdown at scale — detecting and interpreting tables, forms, diagrams, and embedded images, extracting semantic context rather than a flat character stream, and returning bounding boxes for visual elements to support retrieval, grounding, and citation. Trained on business documents across finance, insurance, and scientific research, with support for nine major world languages, an 8,192-token context window, and a ~4.6GB footprint. Priced at $1.50 per 1,000 pages via the Cohere API, with Model Vault for higher-volume/managed deployment and availability on Amazon SageMaker and Microsoft Azure; it can also be deployed privately. Cohere's own comparison places Parse 5 behind larger general-purpose frontier models (GPT-5.5, Opus 4.8, Gemini 3.5 Flash) on raw accuracy, positioning it on price-to-performance and cost per page. Part of Cohere's document-AI line alongside North Micro Vision. Vendor figures, unverified independently at launch.

Cohere's compact document-focused vision-language model, published Aug 12 2026 under Apache 2.0. A 2.4B-parameter VLM combining a custom 400M native-resolution vision encoder, a 2B language model on the Command A+ architecture, and a projector; it preserves aspect ratio for images up to 1654x2339px (an A4 page at 200 dpi). Multilingual visual understanding across documents, charts, and natural images, text output. Vendor-reported: 0.921 DocVQA and 0.808 ChartQA on document tasks, 0.732 RefCOCO on visual grounding, and 0.687 MMBench on general VQA; text-only capability lags larger models. Part of Cohere's North product family alongside North Mini Code. Open weights on Hugging Face.

Cohere's first developer-focused model and the first in its North family of code agents. A 30B-total / 3B-active MoE for agentic coding with a 256K context and up to 64K output, sized to run locally for enterprise coding agents. Apache-2.0.

Enterprise-grade model tuned for RAG, tool use, and multilingual business workloads.

Cohere's smallest, fastest R-series model, tuned for RAG and tool use on modest hardware.

Open multilingual research model covering 23 languages, released by Cohere For AI.

Higher-capability RAG and tool-use model in Cohere's Command R family.

Enterprise RAG-focused model with tool use, citations, multilingual retrieval, and long-context support.