Ofir Press on Nobel-Level Coding AI: Rewriting FFmpeg or SQLite 5x Faster Could Be the Benchmark
OfirPress · x · 2026-10-07
Asked by @DimitrisPapail how to quantify a "Nobel-level" result for coding AI — in tokens, environment complexity, or autonomy — researcher Ofir Press says he's still searching for a good answer. His candidate precursor: an AI rewriting FFmpeg or SQLite to run 5x faster with 2x less memory, using hard performance metrics on foundational software as the yardstick for agentic coding progress.
Related event: Researchers Debate Why AI Cracks Math First, Not Software Engineering(4 posts)→
More from coding & agent
- ThePrimeagen admits agent-driven terminal workflow has taken over his coding — zeeg · 2026-10-08
- ezyang: types are good even for trivially testable things — no need to remember the tests — ezyang · 2026-10-08
- One engineer shipped web, desktop, iOS and Android apps in 6 months with AI — Yuchenj_UW · 2026-10-08
- Google Labs launches a 'GitHub for vibe coders' to store, share and remix AI-built apps — templecrash · 2026-10-08
- Keep coding agents sane: don't cram everything into one thread — msfeldstein · 2026-10-08
- Codex Cloud shipped with day-0 Tailscale support — and Tailscale didn't even know — pvncher · 2026-10-08