HSCodeComp - 4-digit: leaderboard
HS customs code prediction scored at 4-digit (heading) precision, requiring the right product family rather than just the broad chapter.
Metric: Exact Match Accuracy (%). Source: arxiv.org. Saturation forecast: Estimated already saturated. 14 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 69.94 | #152 |
| 2 | GPT-5 | 69.78 | #98 |
| 3 | Claude Sonnet 4 | 62.97 | #199 |
| 4 | Kimi K2 | 62.03 | #243 |
| 5 | DeepSeek R1 | 61.71 | #251 |
| 6 | GPT-4o | 61.55 | #343 |
| 7 | O3 Mini | 56.17 | #272 |
| 8 | DeepSeek V3 | 54.43 | #321 |
| 9 | Qwen 3 235B A22B | 49.53 | #313 |
| 10 | Qwen Max | 48.58 | #382 |
| 11 | Qwen 3 32B | 29.27 | #443 |
| 12 | QwQ-32B | 29.11 | #421 |
| 13 | Qwen 2.5 72B | 12.34 | #499 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=hscodecomp-4-digit · How It Works · Data refreshed daily, snapshot 2026-10-20.