Qwen3.6-27B-FP8

Qwen3.6-27B-FP8 is an open-weight 27-billion-parameter multimodal language model checkpoint from Alibaba's Qwen team, quantized to FP8 precision for efficient inference. It supports text, image, and video inputs with a native 262,144-token context window, designed for agentic coding, reasoning, and long-context multimodal tasks. Source: supabase.