Aya Vision 32B
Aya Vision 32B combines image and text input with multilingual text output. It supports 23 languages and is available through Cohere's API as well as downloadable research weights.
Capabilities and limits
The documented tasks include image captioning, visual questions and translating information from images. It produces text, rather than generating new images. The weight release is subject to noncommercial terms.Aya Vision documentationaya-vision-32b official model card
When this model fits
Practical guidance. This model is worth comparing when the source is visual and the reader needs an answer in another language. Include signs, scanned text and charts in the review set. Ask for the visible evidence behind an answer, and check that the model does not fill unreadable text with a plausible guess. Use Parse instead when the primary goal is converting a large document collection into an ingestion format.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | API, Downloadable |
| Model identifiers | c4ai-aya-vision-32b |
| Context window | 16,000 tokens Published context limit for this API model. Aya Vision documentation |
| Maximum output | 4,000 tokens Published maximum output; request settings and remaining context can reduce it. Aya Vision documentation |
| Model scale | 32 billion parametersAya Vision documentation |
| Weight license | CC-BY-NC 4.0aya-vision-32b official model card |
Related models
- Aya Vision 8B
Cohere · Aya
Eight-billion-parameter research model for image and text understanding across 23 languages.
Launch date not confirmedResearch releaseDownloadable - Command A Vision
Cohere · Command
Vision-language model for documents, charts, tables and visual questions, with up to 20 images per request.
Launch date not confirmedAPI · Downloadable A 2.4-billion-parameter vision-language model for OCR, charts and document questions, with interleaved image and text input.
Launch date not confirmedDownloadable