Vending-Bench Arena - Round 4: leaderboard
Metric: Final money balance ($, mean over runs). Source: andonlabs.com. 4 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.6 | 3030.46 |
| 2 | Gemini 3 Pro | 1235.16 |
| 3 | Claude Opus 4.5 | 1040.28 |
| 4 | GPT-5.2 | 618.9 |
Interactive version: theaggregate.ai/benchmark?slug=vending-bench-arena-round-4 · How It Works · Data refreshed daily, snapshot 2026-09-19.