YC-Bench
Games · 2026-04-01
Collinear AI's long-horizon business simulation: an agent runs a simulated AI startup for one year through a CLI, starting with $200,000, choosing task contracts, assigning eight employees, building client trust and avoiding adversarial clients while payroll grows, with only a 20-round context and a persistent scratchpad. The leaderboard score is final net worth in USD on the default preset, averaged over three seeded worlds with a bankrupt run counted as zero; it is unbounded and heavy-tailed.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Opus 5.5 | 2873725 |
| Grok 4.5 | 2629903 |
| Qwen3.8-Max | 2095838 |
| Kimi K3 | 2053296 |
| Claude Fable 5 | 1977573 |
| Claude Opus 5 | 1911592 |
| MiMo-V2.6-Pro | 1881623 |
| Claude Sonnet 5.5 | 1741315 |
| Opus 4.7 | 1714511 |
| Opus 4.8 | 1711389 |
| Grok 4.7 | 1666389 |
| DeepSeek V4.1 Flash | 1657236 |
| GLM-5.3 | 1597182 |
| Grok 4.6 | 1568241 |
| GLM-5.1 | 1510772 |
| DeepSeek V4 Pro 0813 | 1459743 |
| GPT-6 Sol | 1441793 |
| Qwen3.8-Omni-Flash | 1380061 |
| GPT-6 Astra | 1336466 |
| Opus 4.6 | 1269734 |
| MiMo-V2.5 | 1220880 |
| GLM-5 | 1208190 |
| GPT-5.5 | 1206045 |
| Sonnet 5 | 1163518 |
| GLM-5.3 Flash | 1140949 |
| DeepSeek-V4-Pro | 1066426 |
| GLM-5.2 | 1013158 |
| GPT-5.4 | 1000803 |
| MiniMax M3 | 999506 |
| Gemini 3.5 Flash | 987017 |
| Gemini 3.6 Flash | 968123 |
| GPT-6.1 Sol | 936267 |
| Qwen3.6 Plus (2026-04-02) | 787999 |
| GPT-5.6 Sol | 728001 |
| GPT-5.6 Terra | 636144 |
| Qwen3.7-Max (source unspecified) | 629583 |
| Kimi K2.6 | 511137 |
| Kimi K2.5 | 408822 |
| Gemini 3 Flash Preview | 393736 |
| GPT-6 Luna | 331093 |