Qwen3.5 Small Series Brings Native Multimodal AI to Edge Devices
· Updated
Qwen released the Qwen3.5 Small Model Series - four multimodal models at 0.8B, 2B, 4B, and 9B parameters with native vision from the larger Qwen3.5 family. The 0.8B and 2B target edge devices, the 4B is a lightweight agent base, and the 9B punches well above its weight class.
The benchmark numbers stand out. Qwen3.5-9B scores 80.1 on GPQA Diamond against gpt-oss-120b's 69.0 - outperforming a model 13x larger on graduate-level reasoning. Across IFBench, MMMLU, and MMMU-Pro, the series stays competitive with models 10-30x bigger, including Gemini 2.5 Flash-Lite.
Both instruct and base variants are on Hugging Face and ModelScope. Try the 4B or 9B for edge inference, lightweight agents, or multimodal pipelines where you need vision capability without heavy compute overhead.
🚀 Introducing the Qwen 3.5 Small Model Series
Qwen3.5-0.8B · Qwen3.5-2B · Qwen3.5-4B · Qwen3.5-9B
✨ More intelligence, less compute.
These small models are built on the same Qwen3.5 foundation — native multimodal, improved architecture, scaled RL:
• 0.8B / 2B → tiny, fast, great for edge device
• 4B → a surprisingly strong multimodal base for lightweight agents
• 9B → compact, but already closing the gap with much larger models
And yes — we’re also releasing the Base models as well.
We hope this better supports research, experimentation, and real-world industrial innovation.
Hugging Face: https://t.co/wFMdX5pDjU
ModelScope: https://t.co/9NGXcIdCWI
