🚧 This area is under construction — things may change or break.
Chat with Qwen3.5 on my Orange Pi!
Yes, you can talk to the OPi sitting on my desk. Use this tool to see if this SoC and quant combination fits your use cases.
Note: I only have one OPi Plus, so there may be times when inference will be slow or disabled due to another workload. These other workloads will be noted in the interface.
Hardware
- Board
- Orange Pi 5 Plus
- SoC
- Rockchip RK3588
- RAM
- 32GB LPDDR4X
- CPU
- 4x A76 + 4x A55
Model
- Model
- Qwen3.5-35B-A3B
- Quantization
- IQ4_NL (18.5 GiB)
- Runtime
- ik_llama.cpp
- Reasoning
- Disabled
- Generation
- ~9 tok/sec
- Prompt processing
- ~25 tok/sec
- Context
- 32K tokens