CodeGemma-2B: benchmark results
Provider: Google. Access: API.
Unified ELO 1238 ± 18, rank #2826 of 2928 rated models, from 14 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| EvalPlus (HumanEval+ & MBPP+) | 33.6 | Pass@1 avg (%) | 23.4 |
| Open LLM Leaderboard v1 - GSM8K | 4.62 | Accuracy (%) (5-shot) | 23.1 |
| ShaderMatch | 6.85 | Clone Match Rate (%) | 19 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 41.22 | MC2 (%) (0-shot) | 16.4 |
| Open LLM Leaderboard v1 - MMLU | 28.32 | Accuracy (%) (5-shot) | 15.7 |
| Big Code Memorization - HumanEval pass@1 | 35.37 | HumanEval pass@1 (%) | 14.7 |
| BigCode Models Leaderboard | 27.3 | HumanEval Python Pass@1 (%) | 13.6 |
| Open LLM Leaderboard v1 - HellaSwag | 39.63 | Normalized accuracy (%) (10-shot) | 8 |
| Open LLM Leaderboard v1 - WinoGrande | 53.67 | Accuracy (%) (5-shot) | 8 |
| Big Code Memorization - HumanEval-ET pass@1 | 28.05 | HumanEval-ET pass@1 (%) | 5.9 |
| Open LLM Leaderboard v1 - ARC Challenge | 25.68 | Normalized accuracy (%) (25-shot) | 5 |
| Wolfram LLM Benchmarking Project | 7.7 | Correct Functionality (%) | 3.5 |
Interactive version: theaggregate.ai/model?slug=codegemma-2b · How It Works · Data refreshed daily, snapshot 2026-09-23.