Pixtral 12B
12-billion parameter multimodal model that analyzes images at arbitrary resolutions and aspect ratios.
visionmultimodalmistralopen-source
Pixtral 12B uses a 400M parameter vision encoder to process charts, documents, diagrams, and complex photographs without fixed image downscaling, supporting 128k context lengths.
Similar Tools & Alternatives
Explore other leading solutions in Image Generation.