mlsanityData to 14 Oct 2026

Winning Gold at IMO 2025 with a Model-Agnostic Verification-and-Refinement Pipeline

arXiv 2507.15855 ↗First posted 21 Jul 2025cs.AIcited by 82

Not on the map. Why?

Cited by 48 in LLM reasoning, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

The International Mathematical Olympiad (IMO) is widely regarded as the world championship of high-school mathematics. IMO problems are renowned for their difficulty and novelty, demanding deep insight, creativity, and rigor. Although large language models perform well on many mathematical benchmarks, they often struggle with Olympiad-level problems. Using carefully designed prompts, we construct a model-agnostic, verification-and-refinement pipeline. We demonstrate its effectiveness on the recent IMO 2025, avoiding data contamination for models released before the competition. Equipped with any of the three leading models -- Gemini 2.5 Pro, Grok-4, or GPT-5 -- our pipeline correctly solved 5 out of the 6 problems (≈85.7% accuracy). This is in sharp contrast to their baseline accuracies: 31.6% (Gemini 2.5 Pro), 21.4% (Grok-4), and 38.1% (GPT-5), obtained by selecting the best of 32 candidate solutions. The substantial improvement underscores that the path to advanced AI reasoning requires not only developing more powerful base models but also designing effective methodologies to harness their full potential for complex tasks.

Key facts. Cited by 82 papers in the map (data through 14 Oct 2026). Builds on 17 of its 28 references that are in the map. First posted on arXiv 21 Jul 2025.

Builds on

17 of 28 references.

References · 17

  1. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,459
  2. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,371
  3. 2025Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilitiescited by 3,796
  4. 2023Improving Factuality and Reasoning in Language Models through Multiagent Debatecited by 1,784
  5. 2024OlympiadBench: A Challenging Benchmark for Promoting AGI with Olympiad-Level Bilingual Multimodal Scientific Problemscited by 1,136
  6. 2023Math-Shepherd: Verify and Reinforce LLMs Step-by-step without Human Annotationscited by 928
  7. 2023Language Models can Solve Computer Taskscited by 372
  8. 2025MathArena: Evaluating LLMs on Uncontaminated Math Competitionscited by 297
  9. 2024Large Language Models for Mathematical Reasoning: Progresses and Challengescited by 284
  10. 2024When Can LLMs Actually Correct Their Own Mistakes? A Critical Survey of Self-Correction of LLMscited by 240
  11. 2023REFINER: Reasoning Feedback on Intermediate Representationscited by 205
  12. 2023Baldur: Whole-Proof Generation and Repair with Large Language Modelscited by 119
5 more
  1. 2025Gold-medalist Performance in Solving Olympiad Geometry with AlphaGeometry2cited by 97
  2. 2025Proof or Bluff? Evaluating LLMs on 2025 USA Math Olympiadcited by 74
  3. 2025A Survey on Large Language Models for Mathematical Reasoningcited by 52
  4. 2025A Survey on Data Contamination for Large Language Modelscited by 35
  5. 2025Brains vs. Bytes: Evaluating LLM Proficiency in Olympiad Mathematicscited by 24

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 30; 2026: 52.

82 papers.

Sort:

By area:

  1. 2026Beyond Benchmarks: MathArena as an Evaluation Platform for Mathematics with LLMscited by 137
  2. 2026AutoHarness: improving LLM agents by automatically synthesizing a code harnesscited by 76
  3. 2026Numina-Lean-Agent: An Open and General Agentic Reasoning System for Formal Mathematicscited by 50
  4. 2025Seed-Prover 1.5: Mastering Undergraduate-Level Theorem Proving via Learning from Experiencecited by 44
  5. 2026Towards Autonomous Mathematics Researchcited by 40
  6. 2025Scaling Behaviors of LLM Reinforcement Learning Post-Training: An Empirical Study in Mathematical Reasoningcited by 29
  7. 2026Automated Conjecture Resolution with Formal Verificationcited by 28
  8. 2025MoReBench: Evaluating Procedural and Pluralistic Moral Reasoning in Language Models, More than Outcomescited by 23
  9. 2026InT: Self-Proposed Interventions Enable Credit Assignment in LLM Reasoningcited by 14
  10. 2025Self-Jailbreaking: Language Models Can Reason Themselves Out of Safety Alignment After Benign Reasoning Trainingcited by 12
  11. 2026LEAP: Supercharging LLMs for Formal Mathematics with Agentic Frameworkscited by 12
  12. 2025HERMES: Towards Efficient and Verifiable Mathematical Reasoning in LLMscited by 11
38 more
  1. 2025LSPO: Length-aware Dynamic Sampling for Policy Optimization in LLM Reasoningcited by 9
  2. 2025Semantic Soft Bootstrapping: Long Context Reasoning in LLMs without Reinforcement Learningcited by 9
  3. 2026HalluHard: A Hard Multi-Turn Hallucination Benchmarkcited by 9
  4. 2026Escaping the Cognitive Well: Efficient Competition Math with Off-the-Shelf Modelscited by 9
  5. 2026SorryDB: Can AI Provers Complete Real-World Lean Theorems?cited by 8
  6. 2025PhysicsMinions: Winning Gold Medals in the Latest Physics Olympiads with a Coevolutionary Multimodal Multi-Agent Systemcited by 7
  7. 2025SciAgent: A Unified Multi-Agent System for Generalistic Scientific Reasoningcited by 7
  8. 2025Achieving Olympiad-Level Geometry Large Language Model Agent via Complexity Boosting Reinforcement Learningcited by 7
  9. 2025Multimodal Reasoning for Science: Technical Report and 1st Place Solution to the ICML 2025 SeePhys Challengecited by 6
  10. 2025Population-Evolve: a Parallel Sampling and Evolutionary Method for LLM Math Reasoningcited by 6
  11. 2025Profile-Aware Maneuvering: A Dynamic Multi-Agent System for Robust GAIA Problem Solving by AWorldcited by 5
  12. 2025Off-Trajectory Reasoning: Can LLMs Collaborate on Reasoning Trajectory?cited by 5
  13. 2025CapGeo-Bench: Decoupling Visual Perception from Reasoning and Evaluating Geometric Understandingcited by 5
  14. 2025Honesty over Accuracy: Trustworthy Language Models through Reinforced Hesitationcited by 5
  15. 2026MechMath Agent Team: LLM Driven Agents for Mathematical Researchcited by 5
  16. 2025MarsRL: Advancing Multi-Agent Reasoning System via Reinforcement Learning with Agentic Pipeline Parallelismcited by 4
  17. 2025Intern-S1-MO: Long-horizon Reasoning Agent for Olympiad?Level Mathematical Problem Solvingcited by 4
  18. 2026The Agentic Researcher: A Practical Guide to AI-Assisted Research in Mathematics and Machine Learningcited by 4
  19. 2026Pseudo-Formalization for Automatic Proof Verificationcited by 4
  20. 2025Learning to Reason with Mixture of Tokenscited by 3
  21. 2025Deep Self-Evolving Reasoningcited by 3
  22. 2025Experience-Guided Adaptation of Inference-Time Reasoning Strategiescited by 3
  23. 2025Pessimistic Verification for Open Ended Math Questionscited by 3
  24. 2025IndiMathBench: Autoformalizing Mathematical Reasoning Problems with a Human Touchcited by 3
  25. 2026Learning to Generate Formally Verifiable Step-by-Step Logic Reasoning via Structured Formal Intermediariescited by 3
  26. 2026QED-Nano: Teaching a Tiny Model to Prove Hard Theoremscited by 3
  27. 2026Bolzano: Case Studies in LLM-Assisted Mathematical Researchcited by 3
  28. 2026Agentic Proving for Program Verificationcited by 3
  29. 2026SPIRAL: Learning to Search and Aggregatecited by 3
  30. 2026Parametric Skillscited by 3
  31. 2025Putnam-like dataset summary: LLMs as mathematical competition contestantscited by 2
  32. 2025AsyncSpade: Efficient Test-Time Scaling with Asynchronous Sparse Decodingcited by 2
  33. 2025Butter-Bench: Evaluating LLM Controlled Robots for Practical Intelligencecited by 2
  34. 2026Classroom Final Exam: An Instructor-Tested Reasoning Benchmarkcited by 2
  35. 2026Code2Math: Can Your Code Agent Effectively Evolve Math Problems Through Exploration?cited by 2
  36. 2026Verifiable Reasoning for LLM-based Generative Recommendationcited by 2
  37. 2026Self-Trained Verification for Training- and Test-Time Self-Improvementcited by 2
  38. 2026Faster Synchronous On-Policy RL via Straggler-Aware Group Sizingcited by 2

50 most cited of 82 All 82 (JSON)