InternLM2-7B-Reward: benchmark results

Provider: Shanghai AI Lab. Access: Open.

Unified ELO 1607 ± 49, rank #602 of 2656 rated models, from 11 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
RewardBench Chat99.16Accuracy (%)99.4
RewardBench87.59Score (%)86.7
RewardBench Reasoning94.53Accuracy (%)83.4
RewardBench Safety87.16Accuracy (%)72.5
RewardBench Precise IF40Score (%)71.9
RewardBench Chat Hard69.52Accuracy (%)64.2
RewardBench Ties51.64Score (%)43.2
RewardBench Focus70.51Score (%)43.1
RewardBench Math56.28Score (%)18.6
RewardBench Factuality42.11Accuracy (%)8.2
StoryAlign18.7Average (self-reported)7.1

Interactive version: theaggregate.ai/model?slug=internlm2-7b-reward · How It Works · Data refreshed daily, snapshot 2026-09-19.