DeepSeek

DeepSeek V3 0324

Mar 25, 2025Open weightsModels.devHugging Face

Compare

Intelligence

9.7

#438 of 747Rank 438 of 747 measured models

Coding

21.2

#220 of 298Rank 220 of 298 measured models

Output speed

109 t/s

#169 of 405Rank 169 of 405 measured models

Price/1M

$0.405

#245 of 664Rank 245 of 664 measured models

Context window

128K

#777 of 1207Rank 777 of 1207 measured models

Cost per task

$0.0211

#24 of 196Rank 24 of 196 measured models

DeepSeek V3 0324 scores 9.7 on the Artificial Analysis Intelligence Index, ranking 438th of 747 measured models. The closest models on this index are Ling 2.6 Flash (9.7), GPT-4.5 (Preview) (9.6), and Qwen3 Coder 30B A3B Instruct (9.6). At $0.405 per million tokens (3:1 input/output blended price), it is more expensive than Qwen3 Coder 30B A3B Instruct ($0.30). With an output speed of 109 t/s, it is slower than Qwen3 Coder 30B A3B Instruct (156 t/s).