LLM agents beat top numerical optimizers in 4 of 5 black-box optimization tasks
lulzxdxdxd · reddit · 2026-10-14
A new arXiv paper (2610.12183) benchmarks LLM agents against the best numerical optimizers on black-box optimization:
- LLM agents win in 4 of 5 black-box optimization domains.
- Adding extra numerical tools to the agents does not consistently help — gains are inconsistent across tasks.
- The results suggest LLMs are strong general-purpose optimizers, but tool augmentation is not a reliable upgrade path; when to combine them with classical methods remains an open question.
More from coding & agent
- Anthropic built its own source control after agent-driven merges hit 30,000/day — rickasaurus · 2026-10-14
- With superhuman coding agents, changing habits is still the hardest engineering problem — daniel_mac8 · 2026-10-14
- Robo Use launches: one API for agents across 32 robot embodiments — joemeno · 2026-10-14
- 10 Open-Source GitHub Repos That Give Your AI Agent Memory, from Mem0 (67K Stars) to Engram — victor_explore · 2026-10-14
- Free Bots: A Persistent 3D City Where AI Agents Actually Live, Work, and Build Houses — Daniel_Farinax · 2026-10-14
- Caught in the logs: Codex quietly delegated work to DeepSeek Flash — dotey · 2026-10-14