Swallow - Pre-trained English - GSM8K: leaderboard
From the Swallow LLM Leaderboard (Institute of Science Tokyo) pre-trained English board, which evaluates base models without post-training. GSM8K grade-school math word problems, 4-shot; metric is exact match. Upstream publishes 0-1 scores; shown as 0-100, higher is better.
Metric: Exact Match (%). Source: swallow-llm.github.io. Saturation forecast: Estimated already saturated. 50 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 2.5 72B | 87 | #512 |
| 2 | Qwen3-8B-Base | 85.5 | #626 |
| 3 | Qwen 2.5 7B | 83.2 | #895 |
| 4 | Qwen3-30B-A3B-Base | 82.8 | #597 |
| 5 | Llama 4 Scout Base | 81.1 | #708 |
| 6 | Falcon3-10B-Base | 80.2 | #1187 |
| 7 | gemma-3-27B-pt | 80.1 | #505 |
| 8 | Qwen3-14B-Base | 79.9 | #628 |
| 9 | Llama 3.1 70B | 79.8 | #588 |
| 10 | Qwen 2.5 14B | 79.3 | #716 |
| 11 | Falcon3-7B-Base | 76.6 | #1286 |
| 12 | Gemma 2 27B | 75.7 | #803 |
| 13 | Qwen3-4B-Base | 71.9 | #880 |
| 14 | Qwen 2.5 32B | 71.8 | #544 |
| 15 | gemma-3-12B-pt | 70.3 | #686 |
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-gsm8k · How It Works · Data refreshed daily, snapshot 2026-11-05.