新华网财经·2026-02-16 10:06Group 1 - Alibaba has launched a new generation large model, Qwen3.5-Plus, which features an innovative underlying model architecture and has a total parameter count of 397 billion, with only 17 billion activated. Its performance surpasses the trillion-parameter Qwen3-Max model, with a 60% reduction in deployment memory usage and a maximum inference throughput improvement of up to 19 times [2] - The API pricing for Qwen3.5-Plus is set at 0.8 yuan per million tokens, making it a cost-effective option for developers [2] - The Qwen3.5-Plus model has been integrated into the Qwen app and PC platform, and developers can access it through the Mota community and HuggingFace, or directly via Alibaba Cloud's API service [2] Group 2 - Qwen3.5-Plus is positioned as the latest large language model in the Qwen3.5 series, supporting both text and multimodal tasks [5] - The Qwen3.5-397B-A17B is identified as the flagship large language model of the Qwen3.5 open-source series, also supporting text and multimodal tasks [6] - Both models were quietly launched on the chat.qwen.ai page, indicating a strategic move to enhance Alibaba's offerings in the AI space [2]