Deepseek Coder 6.7B Base — Benchmarks
Benchmark scores for Deepseek Coder 6.7B Base aggregated from public leaderboards, with how it ranks among open models. See hardware requirements for what you need to run it.
Overall rank: #114 of 123 open modelscomposite 20.8/100 across 4 benchmarks in 3 categories · methodology
Knowledge
| Benchmark | Score | Open rank | All models |
|---|---|---|---|
| ARC Challenge | 36.4% | #42 / 51 | #66 / 77 |
| MMLU | 36.4% | #70 / 77 | #124 / 136 |
Math
| Benchmark | Score | Open rank | All models |
|---|---|---|---|
| GSM8K | 21.3% | #51 / 60 | #72 / 93 |
Reasoning
| Benchmark | Score | Open rank | All models |
|---|---|---|---|
| WinoGrande | 57.6% | #40 / 46 | #74 / 80 |
Scores aggregated from public benchmark sources (each linked from the benchmark pages). llmrun does not run these benchmarks.