AI model leaderboard
From writing code to changing repositories: can the model build working software?
3 evaluations·3 organizations·10/17 20:05 updated
Coding leaderboardTOP 30
Ranks reflect shared evidence across public evaluations. Each board or filtered view shows at most 30 models. Filtered results retain their original board ranks and scores.
The consensus index is not accuracy. Tied display scores retain the order established by the full evidence.
How to read this board
Combines public evaluations from multiple organizations. Evaluation coverage varies by model.Prices do not affect rankings, missing evaluations are not scored as zero, and the index is not accuracy.
Read the methodology →About prices
API prices come from provider websites and are shown per million tokens. USD quotes are converted to CNY using exchange rates dated 2026-10-16.Cached prices apply to input cache hits. Cache writes, storage and subscriptions are charged separately.