Command A Vision
Command A Vision accepts text and images to produce text answers. It focuses on document interpretation, charts, tables and visual questions, with support for up to 20 images in a request.
Capabilities and access
Tool use is not supported by this model, and its output is text rather than generated images. The API permits evaluation; Cohere asks users to contact sales for production deployment. The weight release is separately available under noncommercial terms.Command A Vision documentationCommand A Vision model card
When this model fits
Practical guidance. Choose this route when a person needs an answer about a visual document rather than a full document conversion. Ask the model to identify the page, row or chart element that supports the answer. If the next step must call an external tool, the application needs another component to make that decision. Compare with Parse when the main goal is repeatable ingestion of many documents.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | API, Downloadable |
| Model identifiers | command-a-vision-07-2025 |
| Context window | 128,000 tokens Cohere API model limit; input and generated output must fit the endpoint's context rules. Command A Vision documentation |
| Maximum output | 8,000 tokens Maximum generated output reported by Cohere; request limits can reduce this. Command A Vision documentation |
| Supported languages | English, French, German, Italian, Portuguese and SpanishCommand A Vision model card |
| Weight license | CC-BY-NC 4.0Command A Vision model card |
Related models
- Command A+
Cohere · Command
Sparse multimodal model with 218 billion total parameters, 25 billion active parameters, tool use and 48 languages.
Launch date not confirmedAPI · Downloadable - Aya Vision 32B
Cohere · Aya
A 32-billion-parameter vision-language model with image and text input and text output in 23 languages.
Launch date not confirmedAPI · Downloadable Document parsing model that extracts Markdown, reading order, tables, lists, form fields and visual-element locations.
Launch date not confirmedAPI