Qwen3-VL-32B-Instruct

Qwen3-VL is the vision-language model in the Qwen3 series, achieving state-of-the-art(SOTA)performance on various vision-language(VL)benchmarks. The model supports high-resolution image inputs up to the megapixel level and possesses strong capabilities in general visual understanding, multilingual OCR, fine-grained visual grounding, and visual dialogue. As part of the Qwen3 series, it inherits a powerful language foundation, enabling it to understand and execute complex instructions.