LemmaBench: A Live, Research-Level Benchmark to Evaluate LLM Capabilities in Mathematics

• URL: http://arxiv.org/abs/2602.24173v1
• Date: Fri, 27 Feb 2026 16:52:52 GMT
• Title: LemmaBench: A Live, Research-Level Benchmark to Evaluate LLM Capabilities in Mathematics
• Abstract summary: We present a new approach for benchmarking Large Language Model capabilities on research-level mathematics.<n>Existing benchmarks largely rely on static, hand-curated sets of contest or textbook-style problems as proxies for mathematical research.<n>Instead, we establish an updatable benchmark evaluating models directly on the latest research results in mathematics.
• Score: 5.676144562388248
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: We present a new approach for benchmarking Large Language Model (LLM) capabilities on research-level mathematics. Existing benchmarks largely rely on static, hand-curated sets of contest or textbook-style problems as proxies for mathematical research. Instead, we establish an updatable benchmark evaluating models directly on the latest research results in mathematics. This consists of an automatic pipeline that extracts lemmas from arXiv and rewrites them into self-contained statements by making all assumptions and required definitions explicit. It results in a benchmark that can be updated regularly with new problems taken directly from human mathematical research, while previous instances can be used for training without compromising future evaluations. We benchmark current state-of-the-art LLMs, which obtain around 10-15$\%$ accuracy in theorem proving (pass@1) depending on the model, showing that there is currently a large margin of progression for LLMs to reach human-level proving capabilities in a research context.

Related papers

• MathCoPilot: An Interactive System for Human-AI Symbiotic Paradigm of Mathematical Research [62.221184989046826]
  MathCoPilot is a human-in-the-loop system that embodies a new human--AI symbiotic paradigm for mathematical research.<n>MathCoPilot unifies three core capabilities: an interactive proof blueprint that decomposes a proof into navigable steps the human can directly inspect, direct, and refine; automated proving skill orchestration with adaptive knowledge base search and Lean-integrated iterative verification; and topic-driven paper retrieval and automated formalization into a verified Lean knowledge base.
  arXiv  Detail & Related papers  (2026-07-16T05:22:40Z)
• SABER-Math: Automated Benchmark for Information Retrieval Evaluation in Mathematics [4.685475117723465]
  We introduce SABER-Math, the first fully automated benchmark for evaluating mathematical IR without expert annotation.<n>Starting from 283K high-school-level math problems with solutions, SABER-Math builds challenging reranking tasks in three steps.<n>We evaluate lexical retrievers, specialized mathematical retrieval systems, and recent embedding models.
  arXiv  Detail & Related papers  (2026-06-29T07:32:50Z)
• LEAP: Supercharging LLMs for Formal Mathematics with Agentic Frameworks [85.86474267842907]
  Large Language Models (LLMs) exhibit strong informal mathematical reasoning but struggle to generate verifiable proofs in formal languages like Lean.<n>We present LEAP, an agentic framework that enables general-purpose foundation models to achieve state-of-the-art performance on automated formal theorem proving.
  arXiv  Detail & Related papers  (2026-06-02T08:16:42Z)
• Beyond Benchmarks: MathArena as an Evaluation Platform for Mathematics with LLMs [4.559742899048613]
  We build on the original MathArena benchmark by substantially broadening its scope.<n>MathArena now covers a much wider range of tasks, including proof-based competitions, research-level arXiv problems, and formal proof generation in Lean.<n>The strongest model, GPT-5.5, now reaches 98% on the 2026 USA Math Olympiad and 74% on research-level questions.
  arXiv  Detail & Related papers  (2026-05-01T13:56:34Z)
• LiveMathematicianBench: A Live Benchmark for Mathematician-Level Reasoning with Proof Sketches [61.30693283718321]
  We present LiveMathematicianBench, a dynamic multiple-choice benchmark for research-level mathematical reasoning.<n>By grounding evaluation in newly published theorems, it provides a realistic testbed beyond memorized patterns.<n>It employs a proof-sketchguided distractor pipeline that uses high-level proof strategies to construct plausible but invalid answer choices.
  arXiv  Detail & Related papers  (2026-04-02T08:22:17Z)
• EternalMath: A Living Benchmark of Frontier Mathematics that Evolves with Human Discovery [23.517907682810932]
  We propose a fully automated, theorem-grounded pipeline for evaluating frontier mathematical reasoning.<n>The pipeline transforms recent peer-reviewed mathematical literature into executable and verifiable reasoning tasks.<n>Applying this pipeline yields textbfEternalMath, an evolving evaluation suite derived from contemporary research papers.
  arXiv  Detail & Related papers  (2026-01-04T06:40:25Z)
• MathArena: Evaluating LLMs on Uncontaminated Math Competitions [4.655668424508813]
  MathArena is a new benchmark for evaluating large language models (LLMs)<n>It is based on the following key insight: recurring math competitions provide a stream of high-quality, challenging problems.<n>MathArena is also the first benchmark for proof-writing capabilities.
  arXiv  Detail & Related papers  (2025-05-29T09:28:06Z)
• RealMath: A Continuous Benchmark for Evaluating Language Models on Research-Level Mathematics [30.778394290919582]
  Existing benchmarks for evaluating mathematical reasoning in large language models (LLMs) rely primarily on competition problems, formal proofs, or artificially challenging questions.<n>We introduce RealMath, a novel benchmark derived directly from research papers and mathematical forums that assesses LLMs' abilities on authentic mathematical tasks.
  arXiv  Detail & Related papers  (2025-05-18T23:32:46Z)
• Model Utility Law: Evaluating LLMs beyond Performance through Mechanism Interpretable Metric [99.56567010306807]
  Large Language Models (LLMs) have become indispensable across academia, industry, and daily applications.<n>One core challenge of evaluation in the large language model (LLM) era is the generalization issue.<n>We propose Model Utilization Index (MUI), a mechanism interpretability enhanced metric that complements traditional performance scores.
  arXiv  Detail & Related papers  (2025-04-10T04:09:47Z)
• Sample, Don't Search: Rethinking Test-Time Alignment for Language Models [55.2480439325792]
  We introduce QAlign, a new test-time alignment approach.<n>As we scale test-time compute, QAlign converges to sampling from the optimal aligned distribution for each individual prompt.<n>By adopting recent advances in Markov chain Monte Carlo for text generation, our method enables better-aligned outputs without modifying the underlying model or even requiring logit access.
  arXiv  Detail & Related papers  (2025-04-04T00:41:40Z)
• Challenging the Boundaries of Reasoning: An Olympiad-Level Math Benchmark for Large Language Models [86.45058529521258]
  OlymMATH is a novel Olympiad-level mathematical benchmark designed to rigorously test the complex reasoning capabilities of LLMs.<n>OlymMATH features 200 meticulously curated problems, each manually verified and available in parallel English and Chinese versions.
  arXiv  Detail & Related papers  (2025-03-27T11:20:17Z)
• LeanAgent: Lifelong Learning for Formal Theorem Proving [85.39415834798385]
  We present LeanAgent, a novel lifelong learning framework for formal theorem proving.<n>LeanAgent continuously generalizes to and improves on ever-expanding mathematical knowledge.<n>It generates formal proofs for 155 theorems across 23 diverse Lean repositories.
  arXiv  Detail & Related papers  (2024-10-08T17:11:24Z)
• MathHay: An Automated Benchmark for Long-Context Mathematical Reasoning in LLMs [61.74749961334557]
  MathHay is an automated benchmark designed to assess the long-context mathematical reasoning capabilities of LLMs. We conduct extensive experiments on MathHay to assess the long-context mathematical reasoning abilities of eight top-performing models.
  arXiv  Detail & Related papers  (2024-10-07T02:30:07Z)
• Benchmarking Uncertainty Quantification Methods for Large Language Models with LM-Polygraph [83.90988015005934]
  Uncertainty quantification is a key element of machine learning applications.<n>We introduce a novel benchmark that implements a collection of state-of-the-art UQ baselines.<n>We conduct a large-scale empirical investigation of UQ and normalization techniques across eleven tasks, identifying the most effective approaches.
  arXiv  Detail & Related papers  (2024-06-21T20:06:31Z)
• Mathador-LM: A Dynamic Benchmark for Mathematical Reasoning on Large Language Models [34.814875040792344]
  We introduce Mathador-LM, a new benchmark for evaluating the mathematical reasoning on large language models (LLMs) Mathador-LM is inspired by the Mathador game, where the objective is to reach a target number using basic arithmetic operations on a given set of base numbers. We show that, across leading LLMs, we obtain stable average performance while generating benchmark instances emphdynamically, following a target difficulty level.
  arXiv  Detail & Related papers  (2024-06-18T13:02:12Z)
• GSM-Plus: A Comprehensive Benchmark for Evaluating the Robustness of LLMs as Mathematical Problem Solvers [68.77382332826167]
  Large language models (LLMs) have achieved impressive performance across various mathematical reasoning benchmarks. One essential and frequently occurring evidence is that when the math questions are slightly changed, LLMs can behave incorrectly. This motivates us to evaluate the robustness of LLMs' math reasoning capability by testing a wide range of question variations.
  arXiv  Detail & Related papers  (2024-02-29T15:26:14Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.