Neutron Mojo
Preview tensor, quantization, and transformer-inference libraries for Mojo 1.0.
Overview
- Status
- Preview; APIs still expected to move
- Compute
- CPU-first SIMD tensor and inference paths
- Models
- GGUF and SafeTensors loading with several quantization formats
- Toolchain
- Built on Mojo 1.0 (max >= 26.5)
A dedicated GPU backend and hardened HTTP serving runtime are not currently shipped.