StarCoderBase-7B: benchmark results
Provider: BigCode. Access: Open.
Aggregate Rating 1163 ± 32, rank #9807 of 10916 rated models, from 11 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| CanItEdit (Descriptive) | 32.9 | pass@1 (%; share of the 105 hand-written Python editing prob | 50 |
| CanItEdit (Lazy) | 21.95 | pass@1 (%; share of the 105 hand-written Python editing prob | 33.3 |
| Open LLM Leaderboard v1 - GSM8K | 5.46 | Accuracy (%) (5-shot) | 24.2 |
| Open LLM Leaderboard v1 - MMLU | 28.45 | Accuracy (%) (5-shot) | 15.8 |
| BigCode Models Leaderboard | 28.4 | HumanEval Python Pass@1 (%) | 15.3 |
| EvalPlus (HumanEval+ & MBPP+) | 21.3 | Pass@1 avg (%) | 14.5 |
| CRUXEval | 32.2 | Output Prediction pass@1 (%) | 14.3 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 40.46 | MC2 (%) (0-shot) | 14.3 |
| Open LLM Leaderboard v1 - ARC Challenge | 29.86 | Normalized accuracy (%) (25-shot) | 9 |
| Open LLM Leaderboard v1 - HellaSwag | 43.87 | Normalized accuracy (%) (10-shot) | 8.6 |
| Open LLM Leaderboard v1 - WinoGrande | 54.38 | Accuracy (%) (5-shot) | 8.4 |
Interactive version: theaggregate.ai/model?slug=starcoderbase-7b · How It Works · Data refreshed daily, snapshot 2026-11-08.