Atlas

Benchmarks

← All benchmarks

Codeforces Div. 1 - DeepSeek-reported rating

Code

Codeforces rating on Division 1 contest problems as computed by DeepSeek's internal evaluation, reported in the DeepSeek-R1-0528, V3.1, V3.1-Terminus and V3.2-Exp model cards. The contest set and rating procedure are DeepSeek-internal and differ from the Division 2 set used in the R1 paper, so the row is diagnostic.

Top models (higher is better)

ModelScore
DeepSeek-V3.2-Exp2121
DeepSeek-V3.12091
DeepSeek-V3.1-Terminus2046
DeepSeek-R1-05281930
DeepSeek-R11530
Loading Atlas data…