ELYZA-Thinking-1.0-Qwen-32B: benchmark results
Provider: ELYZA. Released 2025-05-01. Access: Open.
Aggregate Rating 1546 ± 30, rank #1656 of 10870 rated models, from 42 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| JP-TL-Bench - English to Japanese - Hard | 9.06 | LT Score (0-10; Hard items only; anchored pairwise Bradley-T | 65.1 |
| Swallow - Post-trained English - HellaSwag | 88.9 | Accuracy (%) | 62.7 |
| JP-TL-Bench - English to Japanese | 8.88 | LT Score (0-10; anchored pairwise Bradley-Terry vs 20 Base S | 61.9 |
| JP-TL-Bench - Japanese to English - Hard | 8.8 | LT Score (0-10; Hard items only; anchored pairwise Bradley-T | 60.3 |
| JP-TL-Bench - English to Japanese - Easy | 8.86 | LT Score (0-10; Easy items only; anchored pairwise Bradley-T | 58.7 |
| JP-TL-Bench - Japanese to English | 8.34 | LT Score (0-10; anchored pairwise Bradley-Terry vs 20 Base S | 57.1 |
| Swallow - Post-trained Japanese - WMT20 Ja-En | 21.6 | BLEU | 56 |
| Swallow - Post-trained Japanese - JamC-QA | 50.3 | Accuracy (%) | 50.7 |
| JP-TL-Bench - Japanese to English - Easy | 7.73 | LT Score (0-10; Easy items only; anchored pairwise Bradley-T | 46 |
| Swallow - Post-trained Japanese - WMT20 En-Ja | 23.7 | BLEU | 45.5 |
| Swallow - Japanese MT-Bench - Math | 94.3 | Judge Score (normalized, %) | 44.8 |
| Swallow - English MT-Bench - Extraction | 77.8 | Judge Score (normalized, %) | 41.8 |
Interactive version: theaggregate.ai/model?slug=elyza-thinking-1-0-qwen-32b · How It Works · Data refreshed daily, snapshot 2026-11-12.