Magicoder-S-CL-7B: benchmark results
Provider: Other. Access: Open.
Unified ELO 1337 ± 22, rank #2554 of 2928 rated models, from 16 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| InfiBench | 52.71 | Score (%) | 82.9 |
| EvoEval Verbose | 65.85 | Pass@1 (%) | 69 |
| EvoEval Concise | 64.02 | Pass@1 (%) | 68 |
| EvalPlus (HumanEval+ & MBPP+) | 63.9 | Pass@1 avg (%) | 65.7 |
| EvoEval Difficult | 27 | Pass@1 (%) | 64 |
| EvoEval Subtle | 58 | Pass@1 (%) | 64 |
| EvoEval | 46.75 | Pass@1 (%) | 62 |
| EvoEval Tool Use | 52 | Pass@1 (%) | 56 |
| EvoEval Creative | 26 | Pass@1 (%) | 54 |
| EvoEval Combine | 11 | Pass@1 (%) | 50 |
| Open LLM Leaderboard v1 - GSM8K | 14.33 | Accuracy (%) (5-shot) | 36.2 |
| Open LLM Leaderboard v1 - ARC Challenge | 43.34 | Normalized accuracy (%) (25-shot) | 19.2 |
Interactive version: theaggregate.ai/model?slug=magicoder-s-cl-7b · How It Works · Data refreshed daily, snapshot 2026-09-23.