Llama-3-Lumimaid-70B-v0.1: benchmark results
A Llama 3 role-play release by NeverSleep, credited to Undi and IkariDev. The card describes a mixture of role-play and general instruction datasets and uses the Llama 3 prompting format. Provider: NeverSleep. Access: Open.
Aggregate Rating 1524 ± 36, rank #2019 of 10916 rated models, from 6 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - GSM8K | 78.85 | Accuracy (%) (5-shot) | 99.6 |
| Open LLM Leaderboard v1 - MMLU | 80.09 | Accuracy (%) (5-shot) | 99.6 |
| Open LLM Leaderboard v1 - WinoGrande | 84.61 | Accuracy (%) (5-shot) | 94 |
| Open LLM Leaderboard v1 - ARC Challenge | 70.9 | Normalized accuracy (%) (25-shot) | 87.3 |
| Open LLM Leaderboard v1 - HellaSwag | 85.9 | Normalized accuracy (%) (10-shot) | 77.4 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 57.92 | MC2 (%) (0-shot) | 71.3 |
Interactive version: theaggregate.ai/model?slug=llama-3-lumimaid-70b-v0-1 · How It Works · Data refreshed daily, snapshot 2026-11-08.