Benchmark Almanac

FrontierMath Erdős

GPT-6.2 Astra 10.3% leads since November 2026
GPT-6.2 Astra 10.3% leads since November 2026
Only in system cards

Models try to prove or disprove 68 open Erdős problems in Lean 4. A result counts only if the Lean checker accepts the full proof.

By late October 2026, 17 of the 68 conjectures had been resolved by Epoch's runs or by others.

Made by
Epoch AI
Released
September 2026
Kind
Research mathematics
Level
Research frontier
Scored
Checked automatically
Answers
Written proof
Tasks
All public
Measures
Capability
Language
English
In and out
Text, Code → Code, Actions
Interaction
Agentic
Environment
Lean
Items from
Written by experts
Refresh
One fixed set
Run by
Anyone
In views
Agents
Strongest model tested
GPT-6.2 Astra #1 on the capability index
Last in a system card
November 2026
Last model added
November 2026
Scores refreshed
November 20th, 2026

In the authors’ words

We introduce FrontierMath Erdős (FME), a benchmark of 68 Erdős problems that are open as of August 2026. To solve a task in FME, AI systems must resolve (prove or disprove) one of the 68 conjectures in the proof assistant Lean. Our 68 problems were selected by the second author among 652 open problems on erdosproblems.com for their mathematical interest and difficulty. AIs have recently resolved several open problems in mathematics, but these demonstrations fall short of a systematic study of AI capabilities. FME evaluates every AI model on the same fixed problems, autonomously and under the same budget. We evaluated five AIs with a budget of $300 per problem. One (GPT-6 Astra) scored 3%, and all others scored 0%.

Abstract of FrontierMath Erdős (arXiv:2609.25050, CC0)
16
Models scored on its own leaderboard
1
System card reports it, from 1 lab

GPT-6.2 Astra leads with 10.3%

Best score per model from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best.

GPT-6.2 Astra leads with 10.3%, claims included

Best score per model from every source, including labs’ own numbers and claims Epoch AI collected from reports and papers. Labs pick their own settings, so a claim can beat an independent run of the same model. Striped bars are claims, and dotted bars are reruns by Epoch AI or another board.

  1. 1 GPT-6.2 Astra max 10.3%
  2. 2 GPT-6.1 Sol max 4.41%
  3. 3 GPT-6 Astra max 2.94%
  4. 4 Claude Opus 5.5 max 2.94%
  5. 5 Gemini 4 Carbon high; task version 0.1.11rc11 2.94%
  6. 6 Claude Sonnet 5.5 max 2.90%
  7. 7 GPT-6 Sol max 1.47%
  8. 8 Gemini 4 Argon high 1.47%
  9. 9 GPT-5.5 xhigh 0%
  10. 10 Claude Fable 5 max 0%
  11. 11 GPT-5.6 Sol max 0%
  12. 12 Kimi K3 max 0%
  13. 13 Claude Opus 5 max 0%
  14. 14 Claude Fable 5.1 max 0%
  15. 15 Gemini 3.8 Flash high 0%
  16. 16 Grok 4.7 xhigh 0%

Source: Official FrontierMath Erdős leaderboard

#ModelScoreSettingReleasedRun onSource
1 GPT-6.2 Astra 10.3% max, Epoch Nov 2026 Oct 2026 Official leaderboard
2 GPT-6.1 Sol 4.41% max, Epoch Sep 2026 Oct 2026 Official leaderboard
3 GPT-6 Astra 2.94% max, Epoch Sep 2026 Aug 2026 Official leaderboard
4 Claude Opus 5.5 2.94% max, Epoch Sep 2026 Sep 2026 Official leaderboard
5 Gemini 4 Carbon 2.94% high; task version 0.1.11rc11, Epoch Oct 2026 Nov 2026 Official leaderboard
6 Claude Sonnet 5.5 2.90% max, Epoch Sep 2026 Oct 2026 Official leaderboard
7 GPT-6 Sol 1.47% max, Epoch Sep 2026 Sep 2026 Official leaderboard
8 Gemini 4 Argon 1.47% high, Epoch Sep 2026 Oct 2026 Official leaderboard
9 GPT-5.5 0% xhigh, Epoch Apr 2026 Aug 2026 Official leaderboard
10 Claude Fable 5 0% max, Epoch Jun 2026 Aug 2026 Official leaderboard
11 GPT-5.6 Sol 0% max, Epoch Jul 2026 Aug 2026 Official leaderboard
12 Kimi K3 0% max, Epoch Jul 2026 Aug 2026 Official leaderboard
13 Claude Opus 5 0% max, Epoch Jul 2026 Aug 2026 Official leaderboard
14 Claude Fable 5.1 0% max, Epoch Sep 2026 Sep 2026 Official leaderboard
15 Gemini 3.8 Flash 0% high, Epoch Sep 2026 Sep 2026 Official leaderboard
16 Grok 4.7 0% xhigh, Epoch Sep 2026 Sep 2026 Official leaderboard
#ModelScoreSettingDateSource
1 GPT-6.2 Astra 10.3% Nov 2026 Lab model card
2 GPT-6.1 Sol 4.41% max, Epoch Sep 2026 Official leaderboard
3 GPT-6 Astra 2.94% max, Epoch Sep 2026 Official leaderboard
4 Claude Opus 5.5 2.94% max, Epoch Sep 2026 Official leaderboard
5 Gemini 4 Carbon 2.94% high; task version 0.1.11rc11, Epoch Oct 2026 Official leaderboard
6 Claude Sonnet 5.5 2.90% max, Epoch Sep 2026 Official leaderboard
7 GPT-6 Sol 1.47% max, Epoch Sep 2026 Official leaderboard
8 Gemini 4 Argon 1.47% high, Epoch Sep 2026 Official leaderboard
9 GPT-5.5 0% xhigh, Epoch Apr 2026 Official leaderboard
10 Claude Fable 5 0% max, Epoch Jun 2026 Official leaderboard
11 GPT-5.6 Sol 0% max, Epoch Jul 2026 Official leaderboard
12 Kimi K3 0% max, Epoch Jul 2026 Official leaderboard
13 Claude Opus 5 0% max, Epoch Jul 2026 Official leaderboard
14 Claude Fable 5.1 0% max, Epoch Sep 2026 Official leaderboard
15 Gemini 3.8 Flash 0% high, Epoch Sep 2026 Official leaderboard
16 Grok 4.7 0% xhigh, Epoch Sep 2026 Official leaderboard

Epoch AI has run 16 models. Labs report scores for 1 model in their cards. Switch the scores above to rank them together.

4 models have held number one on the official leaderboard since its release in September 2026 4 models have held number one with claims included since its release in September 2026

Lab claims count from the day the card came out, independent runs from the model’s release. We draw one row each for the benchmark’s own leaderboard and Epoch AI’s runs.

OpenAI
Own leaderboard 4 leaders
Epoch AI runs 4 leaders

Sources: Official FrontierMath Erdős leaderboard and Epoch AI runs (CC BY 4.0)

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
On the benchmark’s own leaderboard Run by Epoch AI Reported by the lab in its model card Best so far
0% 20% 40% 60% 80% 100% 2026 2027 Near the ceiling (85%) GPT-6.2 Astra GPT-6.1 Sol GPT-5.5 0% 20% 40% 60% 80% 100% 2026 2027 Near the ceiling (85%) GPT-6.2 Astra GPT-6 Astra

Sources: Official FrontierMath Erdős leaderboard, Epoch AI runs (CC BY 4.0) and labs’ model cards

Labs report it only in system cards

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark. Some system cards do, below.

One system card reports it

Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.

DateDocumentScoreAs the document writes itSection
Nov 17, 2026 GPT-6.2 Astra system card 10.3 Epoch AI FrontierMath Erdős, Lean solutions to 7 of 68 problems with a $300 budget per question (2.1.3.2.2) AI R&D risk

One lab claim next to the leaderboard

Each score a lab reported for its own model, beside the same model's best score on the leaderboard. Gaps mostly come from different harnesses, effort levels and task subsets. 1 of them match, which usually means one side copied the other. Every benchmark.

Model Lab says Board says Gap
GPT-6.2 Astra 10.3% Epoch AI FrontierMath Erdős, Lean solutions to 7 of 68 problems with a $300 budget per question (2.1.3.2.2) 10.3% Leaderboard, max, Epoch Same

More in Math

All 140 in Math
  • AIME 2026 In use Models solve the 30 integer-answer problems from the 2026 AIME I and II math competitions. Scored by exact answer match. GPT-5.5 100%
  • IMO-AnswerBench In use 400 olympiad problems with short verifiable answers, robustified against memorization. Scored by answer match. GLM-5.2 91%
  • HMMT 2026 In use Models solve the 33 final-answer problems from the February 2026 Harvard-MIT Mathematics Tournament. Scored by exact answer match. GPT-5.5 98.5%
  • PolyMath In use Models solve 9,000 parallel math problems in 18 languages across four difficulty levels, from school math to frontier math. Final answers are checked. Qwen3.7-Max 86.5%
  • MathArena (Apex, IMO, USAMO) In use Evaluates models on newly released math competitions after release. Final-answer or human/LLM-judged proof scores. Claude Opus 4.8 81.3%
  • MathArena Apex Shortlist In use Models solve 47 hard final-answer problems that MathArena selected from 2025 competitions, where the best models scored about 50% at selection. Scored by accuracy over repeated attempts. GPT-5.5 98.4%

Data audit

All 12 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 5 pass
  • Pass Independent runs Scores from the official leaderboard (16 entries), Epoch AI’s runs (16 models) and system cards (1 score).
  • Pass Models scored 16 models we track.
  • Pass Newest score The newest is from November 2026, for GPT-6.2 Astra.
  • Pass Frontier models 8 of the top 10 have a score, led by GPT-6.2 Astra (#1).
  • Pass Headroom The best is 10.3%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on epoch.ai: 16 rows for models we track at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 2 pass
  • Pass Lab claims The one pair agrees within 10 points.
  • Pass Known issues No published audit found problems.

How we check