Llama-3.3-Swallow-70B-v0.4: benchmark results
Provider: Meta.
Aggregate Rating 1594 ± 39, rank #1064 of 10916 rated models, from 22 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Swallow - Pre-trained Japanese - Average | 62.9 | Average Score (%) | 100 |
| Swallow - Pre-trained Japanese - NIILC | 73.2 | Character F1 (%) | 100 |
| Swallow - Pre-trained Japanese - WMT20 En-Ja | 32.7 | BLEU | 100 |
| Swallow - Pre-trained Japanese - WMT20 Ja-En | 26 | BLEU | 100 |
| Swallow - Pre-trained Japanese - XL-Sum | 28.3 | ROUGE-2 (%) | 100 |
| Swallow - Pre-trained English - HumanEval | 70.9 | Pass@1 (%) | 99 |
| Swallow - Pre-trained English - Average | 71.9 | Average Score (%) | 98 |
| Swallow - Pre-trained English - BBH | 84 | Exact Match (%) | 98 |
| Swallow - Pre-trained English - GSM8K | 86.3 | Exact Match (%) | 98 |
| Swallow - Pre-trained English - TriviaQA | 81.7 | Exact Match (%) | 98 |
| Swallow - Pre-trained Japanese - JCommonsenseQA | 96.7 | Accuracy (%) | 98 |
| Swallow - Pre-trained Japanese - JEMHopQA | 67.1 | Character F1 (%) | 98 |
Interactive version: theaggregate.ai/model?slug=llama-3-3-swallow-70b-v0-4 · How It Works · Data refreshed daily, snapshot 2026-11-08.