Qwen3.5 122B A10B

Roboflow's Playground page describes Qwen3.5-122B-A10B as a high-capacity multimodal MoE model from Alibaba's Qwen team with 122B total parameters and roughly 10B activated per token through sparse expert routing. The model processes both text and visual inputs within a unified multimodal framework, enabling document u The page reports Roboflow's legacy vision evaluation showing a 76.12% pass rate across 67 visual understanding tasks (ranking 9th of 77 models, better than 86%). Playground usage statistics show 41 inferences in the past 30 days with average latency of 18.02s, and speed averaging 1.77s. Interactive demo capabilities in