Alibaba Releases Qwen 3.5 Open-Source Model Series

Alibaba Group released the Qwen 3.5 series, a collection of four open-source multimodal large language models available in 0.8B, 2B, 4B, and 9B parameter sizes. The models are designed to process both text and images, providing developers with advanced reasoning capabilities for logical tasks, mathematics, and document analysis.

The company claims the 9B model delivers performance comparable to much larger systems, such as GPT-oSS-120B, and competes with Gemini and ChatGPT. To support edge computing, Alibaba optimized the smaller 0.8B and 2B versions for portable devices with limited processing power. The weights for the series are hosted on ModelScope and Hugging Face.

Elon Musk, head of xAI, responded to the release by describing the Qwen 3.5 series as having "impressive intelligence density."