GSM-8K (CoT): leaderboard
8,500 grade-school math word problems (7,500 train, 1,000 test) taking 2 to 8 arithmetic steps, from OpenAI (2021), solved with chain-of-thought prompting; exact-match on the final numeric answer.
Metric: Score (%). Source: arxiv.org. Source: benchmarklist.com. Saturation forecast: Estimated already saturated. 5 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Llama 3.1 405B Instruct | 96.8 | #457 |
| 2 | Llama 3.1 70B Instruct | 95.1 | #561 |
| 3 | Llama 3 70B Instruct | 93 | #634 |
| 4 | Llama 3.1 8B Instruct | 84.5 | #1035 |
| 5 | Llama 3 8B Instruct | 80.6 | #1129 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=gsm-8k-cot · How It Works · Data refreshed daily, snapshot 2026-10-20.