Running Lean at Scale

60 points by eab- 6 hours ago on hackernews | 5 comments

auggierose | 5 hours ago

Very interesting. Do I get this right, running 500000 instances for 1 hour can be done for about $5000, or are there many hidden costs? (500000 * $0.01).

RGamma | 5 hours ago

This is part of the work that lead to Aristotle, the system that performed at Gold level at IMO: https://arxiv.org/abs/2510.01346

jarmitage | 3 hours ago

In case you want to try Aristotle, I asked Claude Code to make a plugin for it here https://github.com/afhverjuekki/claude-code-aristotle-plugin
am i understanding it right that this is used to validate the output of llms? any other uses for distributed lean? genuinely curious

UltraSane | an hour ago

Lean is an automated theorem prover. It decides if a given proof is true or not. This uses LLMs to try to write proofs for a given problem