Get the latest tech news

Comparing our Rust-based indexing and querying pipeline to Langchain

Should you use Rust for LLM based tools, and should you do it for performance reasons? Let's take a look at some benchmarks.

We frequently get asked why we’re building text (code) processing software in Rust and not in Python since our performance would be bottlenecked by the LLM inference anyway. Our motivation for using Rust to implement Swiftide is manifold: we want to build a fast and efficient toolchain, and we want to have it be reliable and both easy to write and maintain. Realistically, although Rust boasts performance benefits like zero-cost abstractions and fearless concurrency, that doesn’t mean it’s going to make your project run ten times faster.

Get the Android app