DeepSeek-Coder-7B-Instruct-v1.5: benchmark results
Provider: DeepSeek. Access: Open.
Aggregate Rating 1340 ± 36, rank #7961 of 10916 rated models, from 16 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| EvoEval Combine | 24 | Pass@1 (%) | 86 |
| EvoEval Subtle | 66 | Pass@1 (%) | 85 |
| EvoEval Difficult | 37 | Pass@1 (%) | 81 |
| EvoEval | 53.23 | Pass@1 (%) | 78 |
| EvoEval Tool Use | 60 | Pass@1 (%) | 77 |
| EvoEval Concise | 67.68 | Pass@1 (%) | 75 |
| EvoEval Creative | 37 | Pass@1 (%) | 74 |
| EvalPlus (HumanEval+ & MBPP+) | 66.8 | Pass@1 avg (%) | 72.6 |
| EvoEval Verbose | 65.24 | Pass@1 (%) | 66 |
| Open LLM Leaderboard v1 - GSM8K | 20.39 | Accuracy (%) (5-shot) | 41.3 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 46.73 | MC2 (%) (0-shot) | 35.4 |
| FullStackBench (2024-12 bilingual) | 43.48 | Pass@1 (%), 3,374 problems: 1,687 English and 1,687 Chinese; | 34.6 |
Interactive version: theaggregate.ai/model?slug=deepseek-coder-7b-instruct-v1-5 · How It Works · Data refreshed daily, snapshot 2026-11-08.