Alibaba Launches Qwen3.5 Small Model Series, Bringing Powerful AI to Edge Devices with Minimal Compute

Alibaba launches the Qwen3.5 Small Model Series, a lineup of efficient AI models ranging from 0.8B to 9B parameters built for edge and on-device deployment, with the 9B flagship using Scaled Reinforcement Learning to rival models up to 10x its size and a native multimodal 4B model delivering strong visual reasoning at a fraction of the compute cost.

• Alibaba's Qwen team releases the Qwen3.5 Small Model Series, a family of LLMs ranging from 0.8B to 9B parameters designed for on-device and edge deployment, prioritizing efficiency over raw scale under the philosophy of 'More Intelligence, Less Compute.'
• The 4B model introduces native multimodal architecture, processing text and visual data in a unified latent space rather than relying on adapter-based vision towers, resulting in improved spatial reasoning and OCR accuracy for lightweight agentic applications.
• The flagship 9B model leverages Scaled Reinforcement Learning to optimize logical reasoning paths, closing the performance gap with models 5x to 10x its size, while the 0.8B and 2B variants target ultra-low latency and minimal VRAM usage for mobile and IoT environments.