Qwen3.5-0.8B — RKLLM v1.2.3 (w8a8, RK3588)
RKLLM conversion of Qwen/Qwen3.5-0.8B for Rockchip RK3588 NPU inference.
Converted with RKLLM Toolkit v1.2.3. This is the base model variant — suitable for custom fine-tunes or applications that supply their own system prompting.
Key Details
| Property | Value |
|---|---|
| Base Model | Qwen/Qwen3.5-0.8B |
| Toolkit Version | RKLLM Toolkit v1.2.3 |
| Runtime Version | RKLLM Runtime ≥ v1.2.1 (v1.2.3 recommended) |
| Quantization | w8a8 (8-bit weights, 8-bit activations) |
| Target Platform | RK3588 |
| NPU Cores | 3 |
| Thinking Mode | ❌ Not applicable (base model) |
| Languages | English, Chinese (multilingual) |
Why This Model?
Qwen3.5-0.8B is the successor to Qwen3-0.6B, offering improved architecture and training. At under 1GB after quantisation it is the smallest model in the RK3588 lineup — ideal for low-latency responses, always-on assistants, or edge deployments where memory is tight.
Hardware Tested
- Orange Pi 5 Plus — RK3588, 16GB RAM, Armbian Linux
- RKNPU driver 0.9.8
- RKLLM Runtime v1.2.3
Usage
With the RKLLM API Server
mkdir -p ~/models/Qwen3.5-0.8B
cd ~/models/Qwen3.5-0.8B
git lfs install && git clone https://huggingface.co/GatekeeperZA/Qwen3.5-0.8B-RKLLM-v1.2.3 .
Use with GatekeeperZA/RKLLM-API-Server.
With the official RKLLM API demo
git clone https://github.com/airockchip/rknn-llm.git
cd rknn-llm/examples/rkllm_api_demo
./build/rkllm_api_demo /path/to/Qwen3.5-0.8B-rk3588-w8a8.rkllm 4096 8192
File Listing
| File | Description |
|---|---|
Qwen3.5-0.8B-rk3588-w8a8.rkllm |
Quantized model for RK3588 NPU |
Compatibility Notes
- Minimum runtime: RKLLM Runtime v1.2.1. v1.2.3 recommended.
- RKNPU driver: ≥ 0.9.6
- SoCs: RK3588 / RK3588S. Not compatible with RK3576 without reconversion.
- RAM: ~1.5GB loaded. Runs on any RK3588 board including 8GB variants.
Acknowledgements
- Alibaba Qwen Team for Qwen3.5
- Rockchip / airockchip for the RKLLM toolkit and runtime
- Converted by GatekeeperZA
- Downloads last month
- -
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support