HSCodeComp - 2-digit: leaderboard
Predict the correct Harmonized System customs code for a real e-commerce product from its title and attributes, scored at 2-digit (chapter) precision — the coarsest level, where most models cluster.
Metric: Exact Match Accuracy (%). Source: arxiv.org. Saturation forecast: Estimated already saturated. 14 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 82.59 | #98 |
| 2 | Gemini 2.5 Pro | 80.54 | #152 |
| 3 | Claude Sonnet 4 | 78.8 | #199 |
| 4 | Kimi K2 | 78.01 | #243 |
| 5 | DeepSeek R1 | 77.22 | #251 |
| 6 | O3 Mini | 77.22 | #272 |
| 7 | DeepSeek V3 | 77.06 | #321 |
| 8 | GPT-4o | 75.47 | #343 |
| 9 | Qwen Max | 71.52 | #382 |
| 10 | Qwen 3 235B A22B | 66.93 | #313 |
| 11 | QwQ-32B | 66.77 | #421 |
| 12 | Qwen 3 32B | 64.4 | #443 |
| 13 | Qwen 2.5 72B | 20.73 | #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-2-digit · How It Works · Data refreshed daily, snapshot 2026-10-20.