ASyMOB: New Benchmark Reveals LLMs Struggle With Symbolic Math Generalization
Researchers have introduced ASyMOB, a dataset of 35,368 symbolic math problems designed to distinguish genuine mathematical reasoning from pattern memorization in large language models. The benchmark, accepted at ICML 2026, applies systematic perturbations to seed problems to test whether models can generalize rather than recall. The findings expose significant fragility in most LLMs under minor problem variations, while also identifying a promising frontier for hybrid LLM and computer algebra system approaches.
ASyMOB (Algebraic Symbolic Mathematical Operations Benchmark) is a newly published diagnostic dataset containing 35,368 validated problems spanning integration, limits, differential equations, series, and hypergeometrics. Each seed problem is systematically altered using symbolic, numeric, and equivalence-preserving transformations to probe whether models truly generalize or merely memorize training data. Evaluation results show that most LLMs experience sharp performance drops under even minor perturbations, though a subset of top-performing systems display a qualitatively different robustness profile, described by the authors as a 'regime shift.' Integrated code tools, such as external code execution environments, were found to stabilize performance especially for weaker models. Notably, the study identifies cases where computer algebra systems (CAS) fail but LLMs succeed, and other cases solvable only through a hybrid LLM-CAS pipeline, suggesting complementary strengths. The benchmark is positioned as a principled tool for measuring progress toward verifiable and trustworthy AI in scientific discovery. The work has been accepted at ICML 2026, with the dataset and code publicly available.
What's missing
The mechanism behind the observed 'regime shift' in top models is not explained at the abstract level, and it is uncertain whether the benchmark has been independently validated by parties outside the authoring team.
What different sources said
- arXiv cs.AICenter
ASyMOB: Algebraic Symbolic Mathematical Operations Benchmark
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.