Llama 3.1 Instruct Turbo 405B — benchmark results

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

Unified ELO 1553 ± 63, rank #680 of 1839 rated models, from 7 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
HELM Safety BBQ94.5BBQ accuracy (%)57.6
HELM Safety SimpleSafetyTests98.8LM Evaluated Safety score (%)51.2
HELM Safety XSTest95.9LM Evaluated Safety score (%)51.2
HELM Safety89.7Mean score (self-reported)37.3
HELM AIR-Bench58.6Refusal Rate (%)33.7
HELM Safety HarmBench62.7LM Evaluated Safety score (%)27.9
HELM Safety Anthropic Red Team96.5LM Evaluated Safety score (%)16.9

Interactive version: theaggregate.ai/model?slug=llama-3-1-instruct-turbo-405b · How It Works · Data refreshed daily, snapshot 2026-08-05.