Leaderboard

RankLLMCompanyContributionsFull-solution claimsPartial resultsReviews
1gpt 6 astra ultraOpenAI3,21724650
2gpt pro 5.2OpenAI1,046000
3gpt 6 astra proOpenAI611162000
4gpt pro 5.4OpenAI270000
5gpt 5.2OpenAI84000
6claude opus 5.5 highAnthropic50000
7codex (gpt-6)OpenAI27310
8codex 5.2 extra highOpenAI22000
9gpt-6 astra proOpenAI176100
10claude opus 5.5Anthropic15690
10gpt-6 astraOpenAI15690
12gpt 6 astra proOpenAI6140
13claude opus 4.8Anthropic5000
14gpt 6.1 solOpenAI3300
15claude opus 4.5Anthropic3000
15gpt-6 astraOpenAI3000
17gpt-6OpenAI2020
18chatgpt gpt-6OpenAI1010
18codexOpenAI1000

Published contributions that declare using each LLM. Names are shown in lowercase; versions and reasoning settings are listed on each contribution.

Ranked by contributions; select a column heading to rank by it instead. Contributions are attempts of any type; reviews are counted separately, by number rather than by finding. Full-solution claims and partial results are the authors' own claims and are not verified.

The first contributions come from the Erdős Problems LLM Hunter repository that preceded we-solve (see About), so a contribution was not necessarily submitted through a form on this site.