Llama 3.2 1B Instruct: benchmark results
Instruction-tuned Llama 3.2 1B checkpoint. Provider: Meta. Released 2024-09-25. Access: Open.
Aggregate Rating 1250 ± 6, rank #1539 of 1656 rated models, from 911 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| SLM Trustworthiness - Privacy | 60.21 | TrustLLM privacy accuracy (%; sub-task mean) | 100 |
| SLM Trustworthiness - TrustLLM Overall | 66.11 | TrustLLM score (%; mean of ethics, privacy, robustness and f | 100 |
| UnpredictaBench - RealWorld | 59.09 | KS@100 (%): share of the 30 RealWorld problems (natural-lang | 100 |
| UnpredictaBench - Shuffling | 36.84 | KS@100 (%): share of the 19 Shuffling problems (random order | 92.1 |
| SLM Trustworthiness - Fairness | 70.81 | TrustLLM fairness accuracy (%; sub-task mean) | 90 |
| Enkrypt AI - Jailbreak Risk | 3.22 | Risk Score | 86.6 |
| Thai LLM NLG - xl_sum_tha_seacrowd_t2t (ROUGE1) | 31.06 | ROUGE1 | 83.6 |
| Thai LLM NLG - xl_sum_tha_seacrowd_t2t (ROUGEL) | 21.33 | ROUGEL | 83.6 |
| Enkrypt AI - Toxicity Risk | 1.18 | Risk Score | 82 |
| Polar (US) - Economic Axis | 71.58 | Economic-axis ICAT (0-100), the StereoSet idealized context | 81.1 |
| IndicIFEval-Ground - Gujarati | 65 | Prompt-level loose accuracy (%, 0-100) on the Gujarati promp | 77.8 |
| Polar (US) | 76.97 | Total ICAT (0-100), the StereoSet idealized context associat | 73 |
Interactive version: theaggregate.ai/model?slug=llama-3-2-1b-instruct · How It Works · Data refreshed daily, snapshot 2026-10-20.