deepseek-coder-6.7B-instruct — benchmark results

Provider: DeepSeek. Released 2023-10-29. Access: Open.

Unified ELO 1335 ± 30, rank #1458 of 1776 rated models, from 28 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
Big Code Memorization - HumanEval-ET pass@162.8HumanEval-ET pass@1 (%)100
EvoEval Verbose72.56Pass@1 (%)91
EvoEval Concise71.95Pass@1 (%)89
Big Code Memorization - HumanEval pass@171.34HumanEval pass@1 (%)88.2
Big Code Memorization - HumanEval pass@5064.94HumanEval pass@50 (%)88.2
Big Code Memorization - HumanEval-ET pass@5056.73HumanEval-ET pass@50 (%)88.2
EvoEval Difficult40Pass@1 (%)88
InfiBench53.25Score (%)84.8
EvoEval53.24Pass@1 (%)80
EffiBench - NMU1.9Normalized Memory Usage79.3
EvalPlus (HumanEval+ & MBPP+)68.4Pass@1 avg (%)78.2
EffiBench - NET2.89Normalized Execution Time74.4

Interactive version: theaggregate.ai/model?slug=deepseek-coder-6-7b-instruct · How the rankings work · Data refreshed daily, snapshot 2026-07-22.