frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Open in hackernews

Grok is a surprisingly good automated theorem prover

2•henryrobbins00•7h ago
TL;DR: I'm working on a Python package called OpenATP [1] that provides a common interface to coding agents for automated theorem proving in Lean. In the latest release, I added support for Leanstral 1.5 [2,3], Grok, and Kimi Code [4]. I was surprised to find that Grok has accuracy competitive with Claude Code and Codex at faster wall-clock times and a fraction of the cost. See the rest of this post and the OpenATP docs [5] for more details.

Automated theorem provers take formal statements (in a proof assistant like Lean) and attempt to synthesize a proof. Unsurprisingly, LLMs have become the dominant approach to this task. Many have fine-tuned LLMs specifically for theorem proving, like Mistral's Leanstral 1.5 model. However, a number of recent papers have shown that general-purpose frontier models in a coding harness are competitive with, and sometimes outperform, these specialized models [6,7].

In my own research, I've had a lot of success using Claude Code and Codex as automated theorem provers. As new models and harnesses have been released in the last few months, I've been curious how they would compare, both in accuracy and in time/cost. I ran into two main issues trying to run benchmarks:

- I needed to bill against subscription plans to make it affordable.

- Compiling Lean is RAM intensive and I needed a cheap way to run multiple isolated agents in parallel remotely.

I created an open-source Python package called OpenATP [1] to solve both of these problems. OpenATP automatically forwards agent CLI credentials into a Docker container to bill against subscription plans and it offers a Modal backend to run multiple agents in parallel remotely (Modal offers $30/month in free compute). See the OpenATP docs [5] for all of the supported provers / harnesses and how to set up credentials for each.

The most recent release of OpenATP added support for Leanstral 1.5, Grok, and Kimi Code. I ran some small benchmarks on a subset of the FATE-H and FATE-X datasets [8] to compare them. The Kimi Code subscription rate limits were too restrictive to include it in the benchmark.

On FATE-H, Claude Code and Codex are the fastest at ~10 minutes per proof. They are also significantly more expensive. Aristotle [9] is free and achieves perfect accuracy in 2x the time. Leanstral 1.5 is also free and has similar time and accuracy to Aristotle (the one miss was due to 60 min timeout). DeepSeek is incredibly cheap, but is also the slowest. It's roughly 3x slower than Claude and hits 3 timeouts.

In comes Grok! Grok achieves accuracy comparable to Claude Code and Codex, but at a fraction of the cost and the fastest wall-clock time. The difference on FATE-X, a more challenging dataset, is even more pronounced. Grok is ~50% faster and ~3x cheaper than Claude Code and Codex. On these more challenging problems, the cheapest provers really begin to struggle. DeepSeek was 4x slower than Grok and hit 4 timeouts.

FATE-H

---------------------------------

prover acc. time cost t/o

---------------------------------

claude 10/10 9:50 $2.05 0

codex 9/10 10:00 $2.68 0

grok 10/10 8:46 $0.94 0

deepseek 7/10 28:21 $0.15 3

leanstral 9/10 23:00 free 1

aristotle 10/10 19:39 free 0

---------------------------------

FATE-X

---------------------------------

prover acc. time cost t/o

---------------------------------

claude 8/9 15:43 $3.60 0

codex 9/9 14:27 $3.70 0

grok 9/9 10:42 $1.31 0

deepseek 3/9 41:48 $0.21 4

---------------------------------

[1] https://github.com/henryrobbins/open-atp

[2] https://news.ycombinator.com/item?id=48780801

[3] https://mistral.ai/news/leanstral-1-5/

[4] https://news.ycombinator.com/item?id=48935342

[5] https://open-atp.henryrobbins.com

[6] https://arxiv.org/abs/2601.14027

[7] https://arxiv.org/abs/2602.24273

[8] https://github.com/frenzymath/FATE

[9] https://aristotle.harmonic.fun

Comments

henryrobbins00•7h ago
Some additional details about these results that didn't make it into the main post:

- The OpenATP "standard provers" were used; see docs [5] for model / harness configuration details

- Time and cost are function of effort level, which may lead to unfair comparison across provers

- FATE-X excludes task 10 since claude and grok hit session limits

- FATE-X excludes leanstral and aristotle due to temporary endpoint failures

- Deepseek's FATE-X accuracy is corrected from 2 to 3 due to verifier bug (now fixed)

- 2 FATE-X deepseek misses are sorry-free, but rely on native_decide

- Claude's FATE-X miss is due to a failed delegation to a background subagent

- All costs come from underlying CLI, except codex which uses pricing table

Ask HN: What changed in your life when you started meditating (and how)?

21•dondraper36•4h ago•12 comments

New Inference Server for DGX Spark: large model C4:55-90 tok/s no spec decode

3•medicis123•2h ago•0 comments

Ask HN: What websites do you visit every day?

3•b8•28m ago•5 comments

Ask HN: Why can't Google create monopolies anymore?

3•BeForce•3h ago•2 comments

Build a website for AI pics for dating apps in 20 hours, launched ads, failed

2•MaximTsyg•3h ago•2 comments

Ask HN: US Equivalent of Anabin?

13•xqb64•1w ago•5 comments

AWS: Inaccurate Estimated Billing Data – $1.7 billion

1313•nprateem•5d ago•754 comments

Ask HN: How to get started with a sustainable bootstrapped business of your own?

2•romanovtexas•7h ago•2 comments

Ask HN: Anyone working on practical robotics (cobots) for the home?

3•bredren•7h ago•0 comments

Grok is a surprisingly good automated theorem prover

2•henryrobbins00•7h ago•1 comments

Building an AI-orchestrated publishing workflow for a long-form writing project

2•tmuhlestein•8h ago•0 comments

IMDB now automatically creating user accounts when you are signed into Amazon

6•terminalbraid•9h ago•9 comments

Thanks HN for 15 years of support and helping me find my life's work

832•nicholasjbs•5d ago•107 comments

Ask HN: How do people keep track of organizational knowledge?

26•kadhirvelm•1d ago•22 comments

Ask HN: How to you optimize output while not burning out?

3•gszr•10h ago•4 comments

Ask HN: What Are You Working On? (July 2026)

292•david927•1w ago•1141 comments

Ask HN: Alternative Careers in the Age of AI

9•helpfulmandrill•11h ago•10 comments

Ask HN: Where is NFTs as a technology or investment?

4•Otternonsenz•7h ago•5 comments

You've reached the end!