Aya Vision

Name: Aya Vision
Author: Cohere

CohereOpen WeightsPending Human Review

Aya Vision is Cohere’s multilingual vision-language family with 8B and 32B language tiers. It combines an autoregressive language model, a SigLIP2 vision encoder, and a multimodal adapter to handle image-and-text prompts in 23 languages. Its capabilities include OCR, image captioning, visual questions, translation of text in images, and document understanding. Both variants support a 16K context and generate text rather than images. Research weights use non-commercial terms. Cohere’s current Chat API serves the 32B checkpoint with a 4K output limit; the 8B hosted checkpoint was retired in April 2026.

2025-03-04

8B and 32B language tiers

Autoregressive Transformer with SigLIP2 vision encoder and multimodal adapter

CC-BY-NC-4.0 with acceptable-use addendum

Compare with other models

Specifications

Parameters: 8B and 32B language tiers
Architecture: Autoregressive Transformer with SigLIP2 vision encoder and multimodal adapter
License: CC-BY-NC-4.0 with acceptable-use addendum
Context Window: 16,000 tokens
Max Output: 4,000 tokens
Type: multimodal
Modalities: textimage

Benchmark Scores

Advanced Specifications

Model Family: Aya
API Access: Available
Chat Interface: Available
Multilingual Support: Yes
Variants: aya-vision-8b: 16K context; weights availableaya-vision-32b: 16K context; current hosted API

Capabilities & Limitations

Capabilities: vision23-language supportOCRvisual question answeringimage-text translation
Known Limitations: Downloaded weights are licensed for non-commercial use; commercial deployments require separate terms.Accepts images but does not generate them.The 8B hosted endpoint was retired April 4, 2026; weights remain available.Visual interpretations and translated image text can be incorrect.
Notable Use Cases: multilingual visual assistantsdocument understandingimage captioning research

Resources

Related Models

North Small Translate

Cohere

North Small Translate is a dedicated text translation model supporting more than 50 languages. Its 218B-parameter mixture-of-experts network activates 25B parameters and targets long-form translation with efficient serving. Open weights are available in several precisions under CC BY-NC 4.0, while commercial use requires appropriate licensing. It can be evaluated through Cohere’s Chat API and a public demo. The release date follows the September 9 API changelog; the launch blog followed on September 10. Translation quality claims depend on the specific language pair, domain, and evaluation method.

Typetext

Parameters218B total, 25B active

2026-09-09

Open Weights

Details Compare

Tiny Aya Thinker

Cohere

Tiny Aya Thinker is Cohere Labs’ 3.35B multilingual reasoning research release, published with the Building Multilingual Bridges paper. L2-Thinker produces reasoning in the user’s language across 60 languages, while En-Thinker provides an English-reasoning comparison. The models build on a 32K Tiny Aya base checkpoint. Downloadable research weights and a public demonstration support multilingual reasoning experiments. The paper publication precedes the October 5 explanatory blog; that later blog is not the first release.

North Micro Vision Instruct

Cohere

North Micro Vision Instruct is a compact open-weight vision-language model for captioning, visual questions, grounding, and document understanding. It combines a 2B language model with a 400M vision encoder and preserves image aspect ratios through native-resolution processing. The language backbone supports 128K tokens, but multimodal training and validated operating conditions cover only 8K; this page uses the conservative multimodal limit. Apache-licensed weights make it a foundation for domain-specific fine-tuning and local experimentation. It supports interleaved images and text, multilingual prompts, and text output.