YC-Bench — leaderboard
Long-horizon agentic startup simulation: models run an AI company for 12 months managing tasks, employees, clients, and cash flow. Score = final net worth averaged over 3 seeds.
Metric: Net Worth ($K). Source: collinear-ai.github.io. Status: saturation imminent. 29 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Fable 5 | 1977.6 |
| 2 | Claude Opus 4.7 | 1714.5 |
| 3 | Claude Opus 4.8 | 1711.4 |
| 4 | GLM-5.1 | 1510.8 |
| 5 | Claude Opus 4.6 | 1269.7 |
| 6 | MiMo-V2.5 | 1220.9 |
| 7 | GLM-5 | 1208.2 |
| 8 | GPT-5.5 | 1206 |
| 9 | Claude Sonnet 5 | 1163.5 |
| 10 | DeepSeek V4 Pro | 1066.4 |
| 11 | GLM-5.2 | 1013.2 |
| 12 | GPT-5.4 | 1000.8 |
| 13 | MiniMax-M3 | 999.5 |
| 14 | Gemini 3.5 Flash | 987 |
| 15 | Qwen 3.6 Plus | 788 |
Interactive version: theaggregate.ai/benchmark?slug=yc-bench · How the rankings work · Data refreshed daily, snapshot 2026-07-22.