XNews

Perplexity Open-Sources Lily: Custom Inference Engine for Apple Silicon

Perplexity Open-Sources Lily: Custom Inference Engine for Apple Silicon

Perplexity open-sourced Lily, a Rust-based inference engine optimized for Qwen3.6-35B-A3B on Apple Silicon, achieving 1.23x prefill and 1.35x decode throughput vs MLX-LM, with custom Metal kernels and no PyTorch dependency.

Source posts