Agentic LLMs Can Write Rust Code 2x-20x Faster Than State-of-the-Art Libraries
minimaxir · x · 2026-09-22
Max Woolf publishes a 26-minute deep dive showing that modern agentic LLMs—given the right guardrails and constraints—can iteratively optimize Rust code until it beats current state-of-the-art libraries by 2x-20x depending on domain. His earlier 2025 experiment with "write better code" prompting on Python failed because the instruction was underspecified; with post-Opus 4.5 agentic coding and a Rust+PyO3 approach, the same idea now works, and improvements compound with each frontier model release. He includes both his prompts and benchmark results.
More from coding & agent
- Nous ships official plugin to run Hermes Agent on Claude Pro/Max subscriptions, no API key — Teknium · 2026-09-22
- Open-source jevmail sorts 1,000 Gmail messages locally in 76 seconds for 3 cents — _AustinCalvert_ · 2026-09-22
- Kos raises $12M from 8VC to build AI finance agents for data center invoices — ahelkky · 2026-09-22
- Harrison Chase on decision models: agentic systems are just good engineering around models — Hacubu · 2026-09-22
- LangSmith Integrates Jev for Cheap Large-Scale Trace Mining That Feeds Directly Into Evals — hwchase17 · 2026-09-22
- Shopify partners with Muse for agentic checkout across all stores — firstadopter · 2026-09-22