TradeVerse - HS Code Prediction: leaderboard
Part of TradeVerse (Banerjee and Dasgupta, University of Pisa, 2026), built from World Trade Organization specific trade concern meetings: from the full negotiation history of a concern, the model predicts the Harmonized System chapters of the products under discussion. Micro-F1 in percent over 724 concerns, higher is better; zero-shot at temperature 0.
Metric: Micro-F1 (%; multi-label prediction of the Harmonized System chapters under discussion from the full negotiation history of 724 TBT and SPS specific trade concerns, micro-averaged over predicted and gold chapters; zero-shot, temperature 0, through the Together API). Source: arxiv.org. Saturation forecast: Around February 2027. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GLM-5.2 | 65.63 | #138 |
| 2 | GPT-OSS-120B | 63.23 | #700 |
| 3 | DeepSeek V4 Pro | 63.07 | #188 |
| 4 | Kimi K2.7 Code | 62.91 | #185 |
| 5 | Llama 3.3 70B Instruct | 47.36 | #1357 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=tradeverse-hs-code-prediction · How It Works · Data refreshed daily, snapshot 2026-11-13.