Prism ML · Bonsai
1-bit Bonsai 8B
Prism ML's 8.2B one-bit text model, distributed in a 1.15 GB build with reported iPhone and Mac measurements.
- Parameters
- 8.2B
- Input
- Text
- Output
- Text
- Access
- Open source
- Licence
- Apache 2.0
- Released
- Device evidence
- Measured on device
- Last verified
- Formats
- 1-bit, MLX
- Runtimes
- MLX, llama.cpp
- Platforms
- iOS, Linux, macOS
Prism ML reports about 40 tokens per second on an iPhone 17 Pro. The 1.15 GB figure describes the compressed model artefact rather than total runtime memory.