E-commerce Last Exam - E-commerce: leaderboard
E-commerce Last Exam restricted to shopping episodes: search, compare, cart and checkout against a simulated storefront, scored by mean reward over the episode.
Metric: Mean Reward (%). Source: huggingface.co. Saturation forecast: Around May 2028. 29 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Grok 4.5 | 78.65 | #48 |
| 2 | Kimi K3 | 76.89 | #30 |
| 3 | Claude Opus 5 | 76.04 | #12 |
| 4 | Qwen 3.8 Max | 75.61 | #35 |
| 5 | Claude Opus 4.8 | 75.54 | #37 |
| 6 | GPT-5.6 Sol | 75.46 | #19 |
| 7 | Claude Opus 4.6 | 75.28 | #63 |
| 8 | DeepSeek V4 Flash | 75.26 | #137 |
| 9 | Claude Sonnet 5 | 74.86 | #57 |
| 10 | GPT-5.5 | 74.54 | #31 |
| 11 | GPT-5.6 Terra | 74.33 | #41 |
| 12 | Qwen 3.8 Flash | 74.11 | #50 |
| 13 | MiMo-V2.5-Pro | 74.05 | #158 |
| 14 | Qwen 3.8 27B | 73.93 | #105 |
| 15 | Claude Opus 4.7 | 73.76 | #49 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=e-commerce-last-exam-e-commerce · How It Works · Data refreshed daily, snapshot 2026-10-20.