Vending-Bench: leaderboard
Tests long-term coherence in AI agents managing a simulated vending machine business - ordering, inventory management, pricing. Ranked by mean net worth across runs.
Metric: Net Worth ($). Source: andonlabs.com. Saturation forecast: Not forecast. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Grok 4 | 4694.15 | #181 |
| 2 | Gemini 3 Pro | 4387.93 | #80 |
| 3 | GPT-5 | 3578.9 | #98 |
| 4 | Claude Sonnet 4.5 | 2465.02 | #142 |
| 5 | GPT-5.1 | 2379.88 | #135 |
| 6 | Claude 3.5 Sonnet | 2217.93 | #350 |
| 7 | Claude Opus 4 | 2077.41 | #163 |
| 8 | O3 | 1843.11 | #124 |
| 9 | Gemini 2.5 Pro (Preview 03-25) | 789.34 | #123 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=vending-bench · How It Works · Data refreshed daily, snapshot 2026-10-20.