SAD - Count Tokens (Situating Prompt): 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) | 20 | #193 |
| 2 | O1 Mini (2024-09-12) | 16.07 | #390 |
| 3 | Claude 3 Opus (20240229) | 13.47 | #394 |
| 4 | Claude 3 Haiku (20240307) | 13.13 | #758 |
| 5 | Claude 3 Sonnet (20240229) | 13.13 | #666 |
| 6 | Llama 2 70B Chat | 11.73 | #1223 |
| 7 | Claude 3.5 Sonnet (20240620) | 11.53 | #294 |
| 8 | Claude Instant 1.2 | 11.53 | #1004 |
| 9 | Claude 2.1 | 11.4 | #825 |
| 10 | Llama 2 13B Chat | 9.4 | #1444 |
| 11 | GPT-4 Preview (0125) | 8.67 | #472 |
| 12 | GPT-4o | 8.2 | #335 |
| 13 | GPT-4 (0613) | 8 | #477 |
| 14 | Llama 2 7B Chat | 7.27 | #1481 |
| 15 | GPT-3.5 Turbo (0613) | 6.73 | #1104 |
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-situating-prompt · How It Works · Data refreshed daily, snapshot 2026-10-12.