SAD - Count Tokens: leaderboard
Metric: Token-count accuracy (%, higher is better). Source: github.com. Saturation forecast: Estimated already saturated. 23 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | O1 Preview (2024-09-12) | 19.8 | #193 |
| 2 | O1 Mini (2024-09-12) | 16.4 | #390 |
| 3 | Claude Instant 1.2 | 12.8 | #1004 |
| 4 | Claude 3 Sonnet (20240229) | 12.13 | #666 |
| 5 | Llama 2 70B Chat | 11.27 | #1223 |
| 6 | Claude 3.5 Sonnet (20240620) | 11.2 | #294 |
| 7 | Claude 3 Opus (20240229) | 11.07 | #394 |
| 8 | Llama 2 13B Chat | 10.93 | #1444 |
| 9 | Claude 3 Haiku (20240307) | 10.13 | #758 |
| 10 | Claude 2.1 | 9.93 | #825 |
| 11 | GPT-4o | 8.2 | #335 |
| 12 | GPT-4 (0613) | 7.93 | #477 |
| 13 | GPT-4 Preview (0125) | 7.93 | #472 |
| 14 | Llama 2 7B Chat | 7.8 | #1481 |
| 15 | davinci-002 | 5.73 | #1407 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=sad-count-tokens · How It Works · Data refreshed daily, snapshot 2026-10-12.