Breaking News! Alibaba Opens Source of the New Generation Large Model

K-LinePoet · 2026-02-22T20:35:40+00:00

Alibaba released the new generation large model Qianwen 3.5-Plus on New Year's Eve, with 397 billion parameters, surpassing the performance of Qwen 3-Max with trillions of parameters, while reducing deployment VRAM by 60%. The new model supports multiple languages and significantly enhances the capabilities of AI shopping agents. Developers can access it through Alibaba Cloud and the Modo community.

K-LinePoet

2026-02-22 20:35:40

Abstract generation in progress

On Lunar New Year’s Eve, February 16th, Alibaba Open-Sourced the all-new next-generation large model Qwen3.5-Plus (hereafter referred to as “Qwen 3.5”), achieving a fundamental overhaul of the underlying model architecture.

It is understood that the released Qwen3.5-Plus version has a total of 397 billion parameters, with 17 billion active parameters, outperforming the trillion-parameter Qwen3-Max model. Deployment memory usage has been reduced by 60%, and maximum inference throughput can be increased by up to 19 times.

The API price for Qwen3.5-Plus is as low as 0.8 yuan per million tokens, which is 1/18 of Gemini 3 Pro.

According to securities industry sources, compared to previous versions, Qwen 3.5 is pre-trained on a mix of visual and text tokens, with additional data in Chinese, English, multiple languages, STEM, and reasoning.

On January 15th, Qwen App launched the world’s first consumer-grade AI shopping agent. During the Spring Festival, the Qwen AI shopping agent completed 120 million orders in six days. Industry insiders say that with significantly enhanced agent capabilities, Qwen 3.5 may further expand the imagination space for how Qwen App can assist users in work and daily life.

Since open-sourcing in 2023, Alibaba has released over 400 Qwen models. To meet the needs of AI developers and enterprises in different countries, the Qwen large models continue to evolve: Qwen 3.5 now supports 201 languages, expanding the vocabulary size from 150,000 to 250,000, which can improve coding efficiency for minority languages by up to 60%.

It is reported that the Qwen App and PC versions have immediately integrated the Qwen3.5-Plus model. Developers can download the new model from the Magic Community and HuggingFace, or directly access API services via Alibaba Cloud Bailing. Alibaba will soon continue to open-source different sizes and functionalities of the Qwen 3.5 series models. The more powerful flagship model, Qwen3.5-Max, will also be released soon.

View Original

This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.

2 Likes