Alibaba · Qwen3.5
Qwen3.5-0.8B
Alibaba's compact dense vision-language model with a 262K-token context and an official 4-bit MNN export.
- Parameters
- 800M
- Context
- 262,144 tokens
- Input
- Text, Image
- Output
- Text
- Access
- Open weights
- Licence
- Apache 2.0
- Released
- Device evidence
- No device evidence
- Last verified
- Formats
- Safetensors, MNN
- Runtimes
- Transformers, MNN, llama.cpp, MLX
- Platforms
- Android, Linux, macOS, Windows
Alibaba publishes an MNN conversion with a 470 MB language weight file and a separate 63 MB vision encoder. The release names mobile-capable runtimes but provides no phone measurement for this size.