{"url":"/sota/automated-theorem-proving-on-minif2f-test","task":{"name":"Automated Theorem Proving","url":"/task/automated-theorem-proving","note":null},"dataset":{"name":"miniF2F-test","url":"/dataset/minif2f"},"category":"Miscellaneous","categories":["Miscellaneous"],"category_note":null,"description":"The goal of **Automated Theorem Proving** is to automatically generate a proof, given a conjecture (the target theorem) and a knowledge base of known facts, all expressed in a formal language. Automated Theorem Proving is useful in a wide range of applications, including the verification and synthesis of software and hardware systems.\r\n\r\n\r\n<span class=\"description-source\">Source: [Learning to Prove Theorems by Learning to Generate Theorems ](https://arxiv.org/abs/2002.07019)</span>","description_from":"task","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","rank":"the archive's row order at snapshot; not re-ranked","rows_end_at":"2025-07-28","rows_withheld_as_spam":0,"metric_values":"the archive's strings, untouched"},"metrics":["cumulative","Pass@1","Pass@32","Pass@64","Pass@100","ITP","pass@1024","pass@8192"],"metric_direction":{"note":"inferred from the metric name only (the archive records no direction); null = not inferred, chart draws points only","by_metric":{"cumulative":null,"Pass@1":null,"Pass@32":null,"Pass@64":null,"Pass@100":null,"ITP":null,"pass@1024":null,"pass@8192":null}},"counts":{"rows":29,"rows_with_code":20,"rows_with_paper_page":27,"rows_dated":27,"rows_using_additional_data":7},"rows":[{"rank_in_archive_order":1,"model":"Kimina-Prover-Preview","metrics":{"ITP":"Lean","Pass@1":"52.94","Pass@32":"68.85","cumulative":"80.74","pass@1024":"77.87","pass@8192":"80.74"},"uses_additional_data":true,"paper_date":"2025-04-15","paper":"/paper/kimina-prover-preview-towards-large-formal","paper_url":"https://arxiv.org/abs/2504.11354v1","paper_title":"Kimina-Prover Preview: Towards Large Formal Reasoning Models with Reinforcement Learning","code":"https://github.com/moonshotai/kimina-prover-preview","n_code_links":2,"syntology":{"n_ran":3,"n_unverified":0,"n_samples":3,"n_pointer_only_licence":3}},{"rank_in_archive_order":2,"model":"ProofAug","metrics":{"ITP":"Isabelle","Pass@1":"36.5","Pass@100":"52.5","cumulative":"66.0"},"uses_additional_data":false,"paper_date":"2025-01-30","paper":"/paper/efficient-neural-theorem-proving-via-fine","paper_url":"https://arxiv.org/abs/2501.18310v1","paper_title":"Efficient Neural Theorem Proving via Fine-grained Proof Structure Analysis","code":"https://github.com/haoxiongliu/proofaug","n_code_links":1,"syntology":null},{"rank_in_archive_order":3,"model":"DeepSeek-Prover-V1.5","metrics":{"ITP":"Lean","Pass@32":"50.0","Pass@64":"50.7","cumulative":"63.5"},"uses_additional_data":true,"paper_date":"2024-08-15","paper":"/paper/deepseek-prover-v1-5-harnessing-proof","paper_url":"https://arxiv.org/abs/2408.08152v1","paper_title":"DeepSeek-Prover-V1.5: Harnessing Proof Assistant Feedback for Reinforcement Learning and Monte-Carlo Tree Search","code":"https://github.com/deepseek-ai/deepseek-prover-v1.5","n_code_links":2,"syntology":{"n_ran":4,"n_unverified":6,"n_samples":10,"n_pointer_only_licence":0}},{"rank_in_archive_order":4,"model":"Subgoal-XL","metrics":{"ITP":"Isabelle","Pass@32":"39.3","cumulative":"56.1"},"uses_additional_data":true,"paper_date":"2024-08-20","paper":"/paper/subgoalxl-subgoal-based-expert-learning-for","paper_url":"https://arxiv.org/abs/2408.11172v1","paper_title":"SubgoalXL: Subgoal-based Expert Learning for Theorem Proving","code":"https://github.com/zhaoxlpku/subgoalxl","n_code_links":1,"syntology":null},{"rank_in_archive_order":5,"model":"DeepSeek-Prover","metrics":{"ITP":"Lean","Pass@1":"30.0","Pass@64":"46.3","cumulative":"52.0"},"uses_additional_data":true,"paper_date":"2024-05-23","paper":"/paper/deepseek-prover-advancing-theorem-proving-in","paper_url":"https://arxiv.org/abs/2405.14333v1","paper_title":"DeepSeek-Prover: Advancing Theorem Proving in LLMs through Large-Scale Synthetic Data","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":6,"model":"Lyra + GPT-4","metrics":{"ITP":"Isabelle","Pass@100":"47.1","cumulative":"47.1"},"uses_additional_data":false,"paper_date":"2023-09-27","paper":"/paper/lyra-orchestrating-dual-correction-in","paper_url":"https://arxiv.org/abs/2309.15806v4","paper_title":"Lyra: Orchestrating Dual Correction in Automated Theorem Proving","code":"https://github.com/chuanyang-zheng/lyra-theorem-prover","n_code_links":1,"syntology":{"n_ran":7,"n_unverified":1,"n_samples":8,"n_pointer_only_licence":0}},{"rank_in_archive_order":7,"model":"LEGO-Prover ChatGPT","metrics":{"ITP":"Isabelle","Pass@100":"47.1","cumulative":"47.1"},"uses_additional_data":false,"paper_date":"2023-10-01","paper":"/paper/lego-prover-neural-theorem-proving-with","paper_url":"https://arxiv.org/abs/2310.00656v3","paper_title":"LEGO-Prover: Neural Theorem Proving with Growing Libraries","code":"https://github.com/wiio12/LEGO-Prover","n_code_links":1,"syntology":null},{"rank_in_archive_order":8,"model":"Decomposing the Enigma","metrics":{"ITP":"Isabelle","Pass@100":"45.5","cumulative":"45.5"},"uses_additional_data":false,"paper_date":"2023-05-25","paper":"/paper/decomposing-the-enigma-subgoal-based","paper_url":"https://arxiv.org/abs/2305.16366v1","paper_title":"Decomposing the Enigma: Subgoal-based Demonstration Learning for Formal Theorem Proving","code":"https://github.com/hkunlp/subgoal-theorem-prover","n_code_links":1,"syntology":null},{"rank_in_archive_order":9,"model":"Evariste","metrics":{"ITP":"Lean","Pass@64":"41","cumulative":"41"},"uses_additional_data":true,"paper_date":"2022-05-23","paper":"/paper/hypertree-proof-search-for-neural-theorem","paper_url":"https://arxiv.org/abs/2205.11491v1","paper_title":"HyperTree Proof Search for Neural Theorem Proving","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":10,"model":"Evariste-7d","metrics":{"ITP":"Lean","Pass@64":"40.6","cumulative":"40.6"},"uses_additional_data":false,"paper_date":"2022-05-23","paper":"/paper/hypertree-proof-search-for-neural-theorem","paper_url":"https://arxiv.org/abs/2205.11491v1","paper_title":"HyperTree Proof Search for Neural Theorem Proving","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":11,"model":"Evariste-1d","metrics":{"ITP":"Lean","Pass@64":"38.9","cumulative":"38.9"},"uses_additional_data":false,"paper_date":"2022-05-23","paper":"/paper/hypertree-proof-search-for-neural-theorem","paper_url":"https://arxiv.org/abs/2205.11491v1","paper_title":"HyperTree Proof Search for Neural Theorem Proving","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":12,"model":"DSP (540B Minerva informal)","metrics":{"ITP":"Isabelle","Pass@100":"38.9","cumulative":"38.9"},"uses_additional_data":false,"paper_date":"2022-10-21","paper":"/paper/draft-sketch-and-prove-guiding-formal-theorem","paper_url":"https://arxiv.org/abs/2210.12283v3","paper_title":"Draft, Sketch, and Prove: Guiding Formal Theorem Provers with Informal Proofs","code":"https://github.com/facebookresearch/minif2f","n_code_links":3,"syntology":null},{"rank_in_archive_order":13,"model":"Lean Expert Iteration","metrics":{"ITP":"Lean","Pass@1":"29.6","Pass@32":"34.5","Pass@64":"36.6","cumulative":"36.6"},"uses_additional_data":true,"paper_date":"2022-02-03","paper":"/paper/formal-mathematics-statement-curriculum","paper_url":"https://arxiv.org/abs/2202.01344v1","paper_title":"Formal Mathematics Statement Curriculum Learning","code":"https://github.com/openai/lean-gym","n_code_links":1,"syntology":null},{"rank_in_archive_order":14,"model":"GPT-f","metrics":{"ITP":"Metamath","Pass@64":"36.6","cumulative":"36.6"},"uses_additional_data":false,"paper_date":"2022-05-23","paper":"/paper/hypertree-proof-search-for-neural-theorem","paper_url":"https://arxiv.org/abs/2205.11491v1","paper_title":"HyperTree Proof Search for Neural Theorem Proving","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":15,"model":"Thor + expert iteration on autoformalised theorems","metrics":{"ITP":"Isabelle","Pass@1":"35.2","cumulative":"35.2"},"uses_additional_data":true,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":16,"model":"COPRA + GPT-4-turbo","metrics":{"ITP":"Lean","Pass@1":"30.7","cumulative":"30.7"},"uses_additional_data":false,"paper_date":"2023-10-06","paper":"/paper/a-language-agent-approach-to-formal-theorem","paper_url":"https://arxiv.org/abs/2310.04353v5","paper_title":"An In-Context Learning Agent for Formal Theorem-Proving","code":"https://github.com/trishullab/copra","n_code_links":1,"syntology":null},{"rank_in_archive_order":17,"model":"Thor","metrics":{"ITP":"Isabelle","Pass@1":"29.9","cumulative":"29.9"},"uses_additional_data":false,"paper_date":"2022-05-22","paper":"/paper/thor-wielding-hammers-to-integrate-language","paper_url":"https://arxiv.org/abs/2205.10893v1","paper_title":"Thor: Wielding Hammers to Integrate Language Models and Automated Theorem Provers","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":18,"model":"Lean GPT-f","metrics":{"ITP":"Lean","Pass@1":"24.6","Pass@32":"29.2","cumulative":"29.2"},"uses_additional_data":false,"paper_date":"2021-08-31","paper":"/paper/minif2f-a-cross-system-benchmark-for-formal","paper_url":"https://arxiv.org/abs/2109.00110v2","paper_title":"MiniF2F: a cross-system benchmark for formal Olympiad-level mathematics","code":"https://github.com/openai/minif2f","n_code_links":4,"syntology":{"n_ran":3,"n_unverified":0,"n_samples":3,"n_pointer_only_licence":0}},{"rank_in_archive_order":19,"model":"MMOS-DeepSeekMath-7B","metrics":{"ITP":"Lean","Pass@1":"28.3","cumulative":"28.3"},"uses_additional_data":false,"paper_date":"2024-02-23","paper":"/paper/an-empirical-study-of-data-ability-boundary","paper_url":"https://arxiv.org/abs/2403.00799v1","paper_title":"An Empirical Study of Data Ability Boundary in LLMs' Math Reasoning","code":"https://github.com/cyzhh/MMOS","n_code_links":1,"syntology":{"n_ran":9,"n_unverified":2,"n_samples":11,"n_pointer_only_licence":11}},{"rank_in_archive_order":20,"model":"ReProver","metrics":{"ITP":"Lean","Pass@32":"26.5","cumulative":"26.5"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":21,"model":"LLEMMA-7b","metrics":{"ITP":"Lean","Pass@32":"26.2","cumulative":"26.2"},"uses_additional_data":false,"paper_date":"2023-10-16","paper":"/paper/llemma-an-open-language-model-for-mathematics","paper_url":"https://arxiv.org/abs/2310.10631v3","paper_title":"Llemma: An Open Language Model For Mathematics","code":"https://github.com/eleutherai/gpt-neox","n_code_links":4,"syntology":{"n_ran":6,"n_unverified":2,"n_samples":8,"n_pointer_only_licence":0}},{"rank_in_archive_order":22,"model":"LLEMMA-34b","metrics":{"ITP":"Lean","Pass@32":"25.8","cumulative":"25.8"},"uses_additional_data":false,"paper_date":"2023-10-16","paper":"/paper/llemma-an-open-language-model-for-mathematics","paper_url":"https://arxiv.org/abs/2310.10631v3","paper_title":"Llemma: An Open Language Model For Mathematics","code":"https://github.com/eleutherai/gpt-neox","n_code_links":4,"syntology":{"n_ran":6,"n_unverified":2,"n_samples":8,"n_pointer_only_licence":0}},{"rank_in_archive_order":23,"model":"PACT (reproduced by Thor)","metrics":{"ITP":"Isabelle","Pass@1":"24.6","cumulative":"24.6"},"uses_additional_data":false,"paper_date":"2021-02-11","paper":"/paper/proof-artifact-co-training-for-theorem","paper_url":"https://arxiv.org/abs/2102.06203v2","paper_title":"Proof Artifact Co-training for Theorem Proving with Language Models","code":"https://github.com/wellecks/llmstep","n_code_links":4,"syntology":{"n_ran":3,"n_unverified":1,"n_samples":4,"n_pointer_only_licence":0}},{"rank_in_archive_order":24,"model":"COPRA + GPT-4","metrics":{"ITP":"Lean","Pass@1":"23.3","cumulative":"23.3"},"uses_additional_data":false,"paper_date":"2023-10-06","paper":"/paper/a-language-agent-approach-to-formal-theorem","paper_url":"https://arxiv.org/abs/2310.04353v5","paper_title":"An In-Context Learning Agent for Formal Theorem-Proving","code":"https://github.com/trishullab/copra","n_code_links":1,"syntology":null},{"rank_in_archive_order":25,"model":"Sledgehammer + heuristics","metrics":{"ITP":"Isabelle","Pass@1":"20.9","cumulative":"20.9"},"uses_additional_data":false,"paper_date":"2022-10-21","paper":"/paper/draft-sketch-and-prove-guiding-formal-theorem","paper_url":"https://arxiv.org/abs/2210.12283v3","paper_title":"Draft, Sketch, and Prove: Guiding Formal Theorem Provers with Informal Proofs","code":"https://github.com/facebookresearch/minif2f","n_code_links":3,"syntology":null},{"rank_in_archive_order":26,"model":"Lean tidy","metrics":{"ITP":"Lean","Pass@1":"18","cumulative":"18"},"uses_additional_data":false,"paper_date":"2021-08-31","paper":"/paper/minif2f-a-cross-system-benchmark-for-formal","paper_url":"https://arxiv.org/abs/2109.00110v2","paper_title":"MiniF2F: a cross-system benchmark for formal Olympiad-level mathematics","code":"https://github.com/openai/minif2f","n_code_links":4,"syntology":{"n_ran":3,"n_unverified":0,"n_samples":3,"n_pointer_only_licence":0}},{"rank_in_archive_order":27,"model":"COPRA + GPT-3.5","metrics":{"ITP":"Lean","Pass@1":"11.9","cumulative":"11.9"},"uses_additional_data":false,"paper_date":"2023-10-06","paper":"/paper/a-language-agent-approach-to-formal-theorem","paper_url":"https://arxiv.org/abs/2310.04353v5","paper_title":"An In-Context Learning Agent for Formal Theorem-Proving","code":"https://github.com/trishullab/copra","n_code_links":1,"syntology":null},{"rank_in_archive_order":28,"model":"Sledgehammer","metrics":{"ITP":"Isabelle","Pass@1":"10.4","cumulative":"10.4"},"uses_additional_data":false,"paper_date":"2022-05-22","paper":"/paper/thor-wielding-hammers-to-integrate-language","paper_url":"https://arxiv.org/abs/2205.10893v1","paper_title":"Thor: Wielding Hammers to Integrate Language Models and Automated Theorem Provers","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":29,"model":"Metamath GPT-f","metrics":{"ITP":"Metamath","Pass@1":"1.3","cumulative":"1.6"},"uses_additional_data":false,"paper_date":"2021-08-31","paper":"/paper/minif2f-a-cross-system-benchmark-for-formal","paper_url":"https://arxiv.org/abs/2109.00110v2","paper_title":"MiniF2F: a cross-system benchmark for formal Olympiad-level mathematics","code":"https://github.com/openai/minif2f","n_code_links":4,"syntology":{"n_ran":3,"n_unverified":0,"n_samples":3,"n_pointer_only_licence":0}}],"since_archive":{"claim":"Results that newer papers report for their own method, placed here by Syntology. A model pointed at the cell in the paper's own table; the number was read from that cell and checked against this leaderboard's metric, dataset, split and scale; an independent check that saw this leaderboard's other rows and every other leaderboard on the same dataset accepted it. Not reviewed by the paper's authors or by the archive's editors, and not ranked against the archive rows.","extraction_file_present":true,"measurement":{"test_papers":883,"papers_with_output":881,"judged_true":108,"judged":110,"wilson95_lower":0.9361,"measured_on":"2026-09-24","frozen_commit":"0e3de0df94"},"measurement_note":"blind adjudication of accepted entries on a held-out split of archive papers, rules frozen before the test","coverage":{"sentence":"Syntology has checked 6,264 of the 9,581 papers on this site that are newer than the archive; results from the others appear after they are checked.","complete":false,"papers_newer_than_archive":9581,"papers_checked":6264,"papers_extracted_not_yet_verified":0,"boards_without_verdict":2,"papers_not_yet_extracted":3316},"order":"newest first by month (arXiv date, else the arXiv-id month), then arXiv id descending","columns":[],"entries":[]},"syntology":{"read_at":"2026-09-24T18:15:14+00:00","claim":"Per row: N of M harvested code samples from that row's paper executed on a synthesized fixture; the other M-N are unverified. Not a reproduction of the row's number; not a correctness claim. n_pointer_only_licence counts samples the site points at rather than redistributes (a licence axis, independent of ran/unverified).","rows_with_graph_line":10,"rows_with_any_sample_ran":10,"distinct_papers_with_graph_line":7,"distinct_papers_with_any_sample_ran":7,"samples_over_distinct_papers":{"n_ran":35,"n_unverified":12,"n_samples":47,"n_pointer_only_licence":14,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":47,"n_unverified":14,"n_samples":61,"n_pointer_only_licence":14,"note":"row-weighted: a paper behind several rows is counted once per row; inflated relative to samples_over_distinct_papers by design, kept for readers summing the per-row syntology blocks"}}}