ABEJA-QwQ32B-Reasoning-Japanese-v1.0: benchmark results
Provider: Other.
Aggregate Rating 1611 ± 33, rank #918 of 10916 rated models, from 35 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Swallow - Post-trained Japanese - JamC-QA | 62.5 | Accuracy (%) | 83.6 |
| Swallow - English MT-Bench - Coding | 78.8 | Judge Score (normalized, %) | 79.1 |
| Swallow - Post-trained English - HellaSwag | 90.6 | Accuracy (%) | 76.1 |
| Swallow - Post-trained English - MMLU-Pro | 79.8 | Accuracy (%) | 76.1 |
| Swallow - English MT-Bench - Humanities | 73.6 | Judge Score (normalized, %) | 71.6 |
| Swallow - English MT-Bench - Roleplay | 75.4 | Judge Score (normalized, %) | 70.1 |
| Swallow - English MT-Bench - STEM | 72 | Judge Score (normalized, %) | 69.4 |
| Swallow - English MT-Bench - Average | 78.5 | Judge Score (normalized, %) | 67.2 |
| Swallow - Japanese MT-Bench - Roleplay | 68.6 | Judge Score (normalized, %) | 67.2 |
| Swallow - Japanese MT-Bench - Reasoning | 77.2 | Judge Score (normalized, %) | 65.7 |
| Swallow - Japanese MT-Bench - Average | 72.7 | Judge Score (normalized, %) | 64.2 |
| Swallow - Japanese MT-Bench - Writing | 65.1 | Judge Score (normalized, %) | 64.2 |
Interactive version: theaggregate.ai/model?slug=abeja-qwq32b-reasoning-japanese-v1-0 · How It Works · Data refreshed daily, snapshot 2026-11-08.