Llama 3.1 8B Base RM RB2: benchmark results
Provider: Meta. Access: Open.
Unified ELO 1647 ± 44, rank #331 of 1605 rated models, from 17 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| RewardBench | 84.63 | Score (%) | 83.6 |
| RewardBench Safety | 88.51 | Accuracy (%) | 79 |
| RewardBench Chat Hard | 77.85 | Accuracy (%) | 77.8 |
| Themis-CodeRewardBench - Memory Efficiency | 69.9 | Preference accuracy (%; share of the benchmark's preference | 74.5 |
| RewardBench Focus | 83.23 | Score (%) | 71.9 |
| RewardBench Factuality | 72 | Accuracy (%) | 70.4 |
| Themis-CodeRewardBench - Readability and Maintainability | 70.78 | Preference accuracy (%; share of the benchmark's preference | 70 |
| Themis-CodeRewardBench - Security Hardness | 71.24 | Preference accuracy (%; share of the benchmark's preference | 69.4 |
| Themis-CodeRewardBench | 75.77 | Preference accuracy (%; share of the benchmark's preference | 68 |
| Themis-CodeRewardBench - Functional Correctness | 82.57 | Preference accuracy (%; share of the benchmark's preference | 65.3 |
| Themis-CodeRewardBench - Execution Efficiency | 61.42 | Preference accuracy (%; share of the benchmark's preference | 57.1 |
| RewardBench Reasoning | 78.86 | Accuracy (%) | 48.5 |
Interactive version: theaggregate.ai/model?slug=llama-3-1-8b-base-rm-rb2 · How It Works · Data refreshed daily, snapshot 2026-09-26.