EComStage: leaderboard
Evaluates text-based e-commerce assistance across seven perception, planning and action tasks. The overall score and task results share the main-table evaluation protocol. This series reports using Task Score (%). Source: the original paper result table.
Metric: Task Score (%). Source: arxiv.org. Saturation forecast: Estimated already saturated. 33 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3 235B A22B Instruct | 85.61 | #336 |
| 2 | Qwen 2.5 72B Instruct | 85 | #469 |
| 3 | Qwen 3 235B A22B | 84.43 | #323 |
| 4 | DeepSeek V3 | 84.23 | #329 |
| 5 | Claude Sonnet 4 | 84.21 | #207 |
| 6 | Gemini 2.5 Pro | 84.01 | #159 |
| 7 | Qwen 3 30B A3B Instruct | 83.63 | #487 |
| 8 | GPT-4o | 83.17 | #357 |
| 9 | GLM-4 32B | 82.63 | #472 |
| 10 | InternVL3-14B | 82.4 | #506 |
| 11 | Qwen 3 4B Instruct | 82.26 | #661 |
| 12 | DeepSeek R1 | 82.17 | #262 |
| 13 | Qwen 3 30B A3B | 82.05 | #496 |
| 14 | Qwen 2.5 14B Instruct | 82.04 | #654 |
| 15 | Qwen 2.5 32B Instruct | 82.03 | #509 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=ecomstage · How It Works · Data refreshed daily, snapshot 2026-11-05.