Qwen3.5-27B is a native multi-modal large language model launched by Tongyi Qianwen team, with 27B parameters. The model uses an efficient hybrid architecture combining gated delta network and gated attention, natively supports 256K context length and can be extended to approximately 1 million tokens. The model achieves unified visual language basic capabilities through early fusion training, supports text, image and video understanding, and is used in reasoning, programming, intelligent...
Provider: Qwen
Context window: 256K
Pricing: $0.16 input / $1.44 output per 1M tokens
Capabilities: serverless, json_mode, function_calling, structured_output
