Llama 3 Chat 70B: benchmark results
Provider: Meta. Released 2024-04-18. Access: Open.
Aggregate Rating 1502 ± 22, rank #773 of 1640 rated models, from 42 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| SAD - Self-Recognition (Situating Prompt) | 73.69 | Score (%, higher is better) | 90 |
| SAD - Introspection (Situating Prompt) | 35.81 | Score (%, higher is better) | 85 |
| Enkrypt AI - Jailbreak Risk | 3.88 | Risk Score | 84.4 |
| SAD - Count Tokens (Situating Prompt) | 13.13 | Token-count accuracy (%, higher is better) | 81.8 |
| SAD - Predict Words | 40.56 | Self-prediction score (%, higher is better) | 81.8 |
| SAD - Introspection | 34.81 | Score (%, higher is better) | 80 |
| SAD - Self-Recognition | 65.69 | Score (%, higher is better) | 80 |
| SAD - Stages (Situating Prompt) | 48.44 | Score (%, higher is better) | 80 |
| SAD - Count Tokens | 11.73 | Token-count accuracy (%, higher is better) | 77.3 |
| SAD - Predict Words (Situating Prompt) | 38.56 | Self-prediction score (%, higher is better) | 77.3 |
| SAD - Influence | 65.31 | Score (%, higher is better) | 75 |
| SAD - Mini | 60.64 | Score (%, higher is better) | 75 |
Interactive version: theaggregate.ai/model?slug=llama-3-chat-70b · How It Works · Data refreshed daily, snapshot 2026-11-07.