FFmpeg's Assembly code is the go-to test for whether LLMs can out-code human engineers
sh4dy_0011 · x · 2026-11-08
A jab in the "show me an engineer who writes better code than an LLM" debate: RomainHedouin suggests picking any Assembly file in the FFmpeg repo and trying to have an LLM reproduce it — implying frontier models still fall far short on hand-tuned low-level SIMD code. Assembly makes up about 8.4% of FFmpeg's codebase and is maintained by a handful of elite human experts, making it a popular benchmark for probing LLM coding limits.
More from coding & agent
- Backend engineer skill tree meme: prompt engineering now sits on top, languages at the bottom — _jaydeepkarale · 2026-11-08
- What I Learned Building an AI Software Factory: Hands-On Lessons from Weeks of Real Projects — bibryam · 2026-11-08
- Tess creator open-sources fine-tuning recipe: turn Claude Code sessions into a local model — udmrzn · 2026-11-08
- AFK Pilot extension brings Grok, Claude, Codex, DeepSeek and more into one VS Code window — PawelHuryn · 2026-11-08
- ComfyUI workflow: MiniMax H4 Ripple with Pika Flex lets you edit one frame and change the whole video — Maleficent-Tell-2718 · 2026-11-08
- 17 open-source offensive AI security frameworks: Cairn hits 4.9k stars — evilsocket · 2026-11-08