An Out-of-Core LLM Inference Engine Written from Scratch in Rust

Vage91 has published Kortex, an out-of-core large language model inference engine written from scratch in Rust. The project targets inference for models that cannot fit entirely in available GPU or system memory by managing execution and data movement across storage or memory tiers. The supplied announcement does not provide benchmark results, supported model formats, hardware requirements, or implementation details. Its significance is therefore architectural and exploratory at this stage rather than evidence of production readiness. The Hacker News listing records a score of 4 and zero comments.