Pythagoras-Prover: New Efficient AI System Advances Formal Mathematical Proof Generation
Researchers have released Pythagoras-Prover, a family of open-source Lean theorem-proving AI models that achieve state-of-the-art results while using far fewer computational resources than leading competitors. The system introduces curriculum-based training on verified proofs, a dynamic filtering scheme, and a novel data augmentation technique called Augmented Lean Formalisation (ALF) to overcome the scarcity of verified proof data. The work demonstrates that strong formal reasoning performance can be achieved at practical compute budgets, potentially broadening access to automated theorem proving.
Pythagoras-Prover is a compute-efficient, open-source family of Lean theorem provers developed to address two core bottlenecks in formal proof generation: scarce verified training data and the high cost of long reasoning traces during proof search. The family includes autoregressive models at 4B and 32B parameters, as well as a proof-of-concept diffusion-based 4B model that iteratively refines proofs at inference time. Training efficiency is achieved through a curriculum of easy, medium, and hard problems, combined with a dynamic filtering scheme that retains informative proof traces within an 8,000-token context limit. The Augmented Lean Formalisation (ALF) technique generates variants of formal statements via self-distillation, expanding the training corpus without requiring full formal verification of every new instance. Empirically, the 4B model outperforms DeepSeek-Prover-V2-671B on MiniF2F-Test at pass@32 (86.1% vs. 82.4%) using roughly 167 times fewer parameters, while the 32B model sets a new open-source record at 93.0% on MiniF2F-Test and solves 93 of 672 PutnamBench problems. The team also releases MiniF2F-ALF, a contamination-sensitive benchmark on which all evaluated models lose accuracy, with the 32B model remaining the strongest performer.
What's missing
The paper does not report wall-clock training or inference time costs in concrete terms, making it difficult to independently assess the claimed compute efficiency. The diffusion-based 4B prover is described only as a proof-of-concept, and its benchmark performance relative to the autoregressive models is not detailed in the abstract. Additionally, the self-distillation step in ALF may introduce subtle distributional biases that are not fully characterized, and long-term generalization beyond MiniF2F and PutnamBench benchmarks remains an open question.
What different sources said
- arXiv cs.AICenter
Pythagoras-Prover: Advancing Efficient Formal Proving via Augmented Lean Formalisation
Related
Gut Bacteria Enzyme Found to Break Down Heat-Processed Food Compounds, Producing Novel Biogenic Amines
Researchers have discovered that an enzyme in common gut bacteria can degrade N-epsilon-carboxymethyllysine (CML), a compound formed during thermal food processing, producing previously unknown biogenic amines. The enzyme, ornithine decarboxylase SpeC from enterobacteria, acts on CML and related modified lysine derivatives through a low-level 'underground' catalytic activity. This finding suggests a previously unrecognized communication axis between thermally processed dietary compounds and gut microbial physiology, with potential implications for host health.
Full-Length Gene Sequencing Reveals Two Distinct Bacterial Communities in Black-Legged Ticks Expanding Into Canada
Researchers used Oxford Nanopore full-length 16S rRNA gene sequencing to characterize the microbiome of Ixodes scapularis black-legged ticks collected in Nova Scotia, Canada, distinguishing between tick-adapted bacteria and environmentally acquired bacteria. The study comes as I. scapularis — the primary vector of Lyme disease — is rapidly expanding northward into Canada due to climate change. The findings suggest that environmentally derived bacteria in tick microbiomes are not mere contamination, which has implications for how tick microbiome data is collected and interpreted across surveillance studies.
Study Identifies Metabolic Link Between Cell Envelope Stress and Biofilm Formation in Bacteria
Researchers have discovered that the metabolite acetyl-CoA directly inhibits enzymes that degrade the bacterial signaling molecule c-di-GMP, connecting cell envelope biosynthesis stress to biofilm formation in Pseudomonas aeruginosa. The study found that sub-inhibitory concentrations of antibiotics targeting early peptidoglycan biosynthesis — but not other antibiotic classes — elevate c-di-GMP levels by reducing phosphodiesterase activity, with acetyl-CoA competing for the enzyme active site. Because the relevant enzyme domain is broadly conserved across bacterial species, this checkpoint mechanism may be widespread and could have implications for understanding antibiotic-induced biofilm responses.