Mistral has launched the Pixtral 12B, its first multimodal AI model adept at processing both text and image inputs, enhancing interactive capabilities for users.
The Pixtral 12B boasts 12 billion parameters, drawing on Mistral’s Nemo 12B text base, enabling it to respond to queries about images and provide detailed information.
While generative AI is often trained on copyrighted content, leading to legal disputes, Mistral hasn’t disclosed the specific image data leveraged for their new model.
Many generative AI models face copyright challenges as they're built partially on copyrighted material, raising questions about fair use and intellectual property in AI.
[
Collection
]
[
|
...
]