Swallow - Pre-trained English - Average: leaderboard
From the Swallow LLM Leaderboard (Institute of Science Tokyo) pre-trained English board, which evaluates base models without post-training. Unweighted mean of OpenBookQA, TriviaQA, HellaSwag, SQuAD2, XWINO, MMLU, GSM8K, MATH, BBH and HumanEval. Upstream publishes 0-1 scores; shown as 0-100, higher is better.
Metric: Average Score (%). Source: swallow-llm.github.io. Saturation forecast: Estimated already saturated. 50 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 2.5 72B | 72.1 | #512 |
| 2 | Qwen3-30B-A3B-Base | 69.9 | #597 |
| 3 | Qwen3-14B-Base | 69.2 | #628 |
| 4 | Llama 3.1 70B | 68.7 | #588 |
| 5 | Qwen 2.5 32B | 68 | #544 |
| 6 | Qwen3-8B-Base | 67.9 | #626 |
| 7 | gemma-3-27B-pt | 67.5 | #505 |
| 8 | Qwen 2.5 14B | 66.6 | #716 |
| 9 | Gemma 2 27B | 65.6 | #803 |
| 10 | Qwen 2.5 7B | 64.3 | #895 |
| 11 | Falcon3-10B-Base | 63.8 | #1187 |
| 12 | Qwen3-4B-Base | 62.2 | #880 |
| 13 | gemma-3-12B-pt | 61.9 | #686 |
| 14 | Llama 4 Scout Base | 60.6 | #708 |
| 15 | Gemma 2 9B | 59.7 | #924 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=swallow-pre-trained-english-average · How It Works · Data refreshed daily, snapshot 2026-11-05.