Llama 3.1 Instruct Turbo 70B — benchmark results

Provider: Meta. Released 2024-07-23. Access: Open.

Unified ELO 1421 ± 72, rank #1138 of 1839 rated models, from 7 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
HELM Safety BBQ95.4BBQ accuracy (%)68.6
HELM Safety XSTest94.5LM Evaluated Safety score (%)34.9
HELM Safety SimpleSafetyTests92.5LM Evaluated Safety score (%)12.8
HELM Safety84.5Mean score (self-reported)12.3
HELM AIR-Bench42.5Refusal Rate (%)9.3
HELM Safety Anthropic Red Team93.2LM Evaluated Safety score (%)8.1
HELM Safety HarmBench46.9LM Evaluated Safety score (%)8.1

Interactive version: theaggregate.ai/model?slug=llama-3-1-instruct-turbo-70b · How It Works · Data refreshed daily, snapshot 2026-08-05.