DeepSeek Coder v2 Lite Base — Benchmarks

Benchmark scores for DeepSeek Coder v2 Lite Base aggregated from public leaderboards, with how it ranks among open models. See hardware requirements for what you need to run it.

Overall rank: #46 of 123 open modelscomposite 52.3/100 across 5 benchmarks in 3 categories · methodology

Knowledge

Benchmark	Score	Open rank	All models
MMLU-Pro	34.4%	#90 / 119	#216 / 259
ARC Challenge	57.3%	#25 / 51	#37 / 77
MMLU	60.5%	#42 / 77	#88 / 136

Math

Benchmark	Score	Open rank	All models
GSM8K	67.1%	#20 / 60	#29 / 93

Reasoning

Benchmark	Score	Open rank	All models
WinoGrande	72.9%	#20 / 46	#43 / 80

Scores aggregated from public benchmark sources (each linked from the benchmark pages). llmrun does not run these benchmarks.