ERPBench (Market Ecologies) - Arena: leaderboard
ERPBench (2026) evaluates enterprise decision agents in a six-round coupled ERP simulation over 100 fixed problems. This board is the Arena ecology, where the six evaluated agents compete against each other in a shared market; the metric is mean sixth-round company valuation in millions, an unbounded monetary outcome, higher is better. It is a different benchmark from ERP-Bench and from the ERPNext-based ERPBench.
Metric: Mean terminal company valuation (M; sixth-round valuation over 100 ERP problems, six evaluated agents competing in a shared market). Source: arxiv.org. Saturation forecast: Around July 2027. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 263.95 | #91 |
| 2 | DeepSeek V4 Flash | 234.71 | #254 |
| 3 | GPT-5.5 | 194.82 | #48 |
| 4 | Seed 2.0 Pro | 188.17 | #209 |
| 5 | Qwen 3.7 Max | 187.13 | #142 |
| 6 | Claude Opus 4.6 | 173.95 | #100 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=erpbench-market-ecologies-arena · How It Works · Data refreshed daily, snapshot 2026-11-12.