DeepHermes 3 - Llama-3.1 8B Preview (Non-reasoning) — benchmark results

Provider: Nous Research. Released 2025-01-15. Access: Open.

Unified ELO 1228 ± 84, rank #1714 of 1841 rated models, from 7 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
AA Humanity's Last Exam4.26Accuracy (%)17.7
Epoch AI - Scicode9.14Score8.8
AA LiveCodeBench8.47Pass@1 (%)6.1
AA MMLU-Pro36.53Accuracy (%)5.2
Artificial Analysis Intelligence Index2.29Intelligence Index4
AA GPQA Diamond26.97Accuracy (%)3.8
AA MATH-50021.8Accuracy (%)2.9

Interactive version: theaggregate.ai/model?slug=deephermes-3-llama-3-1-8b-preview-non-reasoning · How It Works · Data refreshed daily, snapshot 2026-07-25.