@rohanpaul_ai: New Meta, NYU paper shows that a simple loop with good Lean tools and more compute matches or beats hand-built theorem-…
Summary
A new Meta and NYU paper shows that h1, a simple agent loop giving models fast Lean feedback and search tools, matches or beats hand-designed theorem-proving workflows: it outperformed Goedel-Architect on PutnamBench (598 vs 597 problems) at lower cost, and with Opus 4.8 solved 83/100 hard algebra problems versus 63 for Claude Code.
View Cached Full Text
Cached at: 11/01/26, 07:57 PM
New Meta, NYU paper shows that a simple loop with good Lean tools and more compute matches or beats hand-built theorem-proving workflows.
Systems that write math proofs in Lean, a language a computer can check, increasingly use hand-designed workflows that plan and split up problems. Their gains often come with stronger models or more compute, hiding what the workflow adds.
They built h1, a loop that gives the model fast Lean feedback and search tools, and keeps it working after each failed check. It imposes no plan, only a progress summary before each new round.
With the same model, h1 beat the hand-built Goedel-Architect on PutnamBench, 598 problems to 597, at $1.08 against $1.47 per problem. The interface still matters: with Opus 4.8, h1 solved 83 of 100 hard algebra problems, against 63 for Claude Code.
Similar Articles
@rohanpaul_ai: Another great paper from Google. Shows general LLMs can solve formal math by planning proofs and checking each step. Ra…
A new Google paper introduces LEAP, an agentic framework that enables general LLMs to solve formal math problems by planning proofs and checking each step, raising performance from under 10% to 70% on the Lean IMO benchmark and solving all 2025 Putnam problems.
Researchers are already significantly improving on OpenAI’s recent math results, verified in Lean
Researchers are quickly surpassing OpenAI's recent AI-generated math results, with improved proofs formally verified in Lean, highlighting rapid progress in AI-assisted theorem proving.
h1: A Taste of the Bitter Lesson in Automated Theorem Proving
A paper showing that a simple agentic harness (h1) with an efficient Lean 4 feedback interface achieves state-of-the-art formal theorem proving results on PutnamBench, FATE-X, and FormalConjectures, matching or beating specialized multi-step workflows and settling eight open problems. The authors argue that execution interfaces matter more than complex orchestration and warn against prescribing structured workflows before evaluating frontier models with efficient feedback.
@rohanpaul_ai: Meta paper shows that coding agents get much better when they reuse short summaries of past attempts instead of raw log…
A Meta paper shows that coding agents improve significantly when they reuse short summaries of past attempts instead of raw logs, achieving strong gains on SWE-Bench and Terminal-Bench with Claude 4.5 Opus.
@rohanpaul_ai: Brilliant new paper from Meta, CMU and other labs. Shows that coding agents improve faster by manufacturing their own s…
A new paper from Meta, CMU, and other labs presents Self-play SWE-RL, a method where coding agents train themselves by manufacturing and fixing bugs in real codebases, achieving significant gains on SWE-bench benchmarks without relying on human-written tasks.