A multimodal model that accepts both text and image inputs, built on Alibaba's Qwen 3 architecture at 27 billion parameters. The 'uncensored' label signals that safety filters have been removed or loosened compared to the base release, meaning it will engage with content the original model might decline — a trade-off that shifts responsibility to the deployer. It processes visual and textual information together, inheriting the Qwen 3 family's general reasoning and language capabilities.