RSI-Exam Benchmark: Quantifying Recursive Self-Improvement in LLMs
cihangxie · x · 2026-08-27
RSI-Exam is a new benchmark designed to quantify Recursive Self-Improvement (RSI) capabilities in LLMs. Instead of measuring weight modifications, it tests if an agent can improve an 'executable work product' through hours of autonomous experimentation and verify generalization on a hidden test set. The benchmark includes 88 tasks (35 public, 53 private) spanning 6 domains like physical sciences, AI models, and systems. The leaderboard shows Claude and GPT forming a top tier, though there remains significant room for improvement.
More from coding & agent
- Perplexity open-sources SPACE, the Firecracker microVM sandbox behind Perplexity Computer — perplexity_ai · 2026-10-16
- Blackstone goes all-in on Factory for autonomous software development — matanSF · 2026-10-16
- AI software factories hinge on the Factory Loop, not faster code generation — Pavan_Belagatti · 2026-10-16
- 'Dropping a 3,000-line PR twenty minutes after ticket assignment' — AI coding meme goes viral — facontidavide · 2026-10-16
- Self Labs launches Agent Vault to let AI agents log in and pay with credit cards — EXM7777 · 2026-10-16
- Plane Agents: triggers, not prompts, are what turn chatbots into workflow agents — JosephJacks_ · 2026-10-16