GPT-OSS-Swallow-20B-SFT-v0.1: benchmark results
Provider: OpenAI. Access: API.
Aggregate Rating 1583 ± 36, rank #1217 of 10916 rated models, from 35 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Swallow - Japanese MT-Bench - Coding | 80.9 | Judge Score (normalized, %) | 84.3 |
| Swallow - Japanese MT-Bench - Reasoning | 77.7 | Judge Score (normalized, %) | 73.1 |
| Swallow - Post-trained English - LiveCodeBench | 58.9 | Pass@1 (%) | 73.1 |
| Swallow - Japanese MT-Bench - Math | 97.6 | Judge Score (normalized, %) | 70.9 |
| Swallow - Post-trained English - IFBench | 48.8 | Instruction-Level Strict Accuracy (%) | 70.1 |
| Swallow - English MT-Bench - Reasoning | 84.1 | Judge Score (normalized, %) | 67.2 |
| Swallow - Japanese MT-Bench - Average | 73.1 | Judge Score (normalized, %) | 65.7 |
| Swallow - Post-trained English - GPQA Diamond | 64.8 | Accuracy (%) | 65.7 |
| Swallow - Post-trained English - Average | 70.5 | Average Score (%) | 64.2 |
| Swallow - Post-trained Japanese - WMT20 En-Ja | 24.6 | BLEU | 64.2 |
| Swallow - Post-trained Japanese - JamC-QA | 52.1 | Accuracy (%) | 63.4 |
| Swallow - Japanese MT-Bench - Roleplay | 68 | Judge Score (normalized, %) | 62.7 |
Interactive version: theaggregate.ai/model?slug=gpt-oss-swallow-20b-sft-v0-1 · How It Works · Data refreshed daily, snapshot 2026-11-08.