OptiVerse - Game Optimization: leaderboard
Metric: Accuracy (%) on the 51 game optimization problems of OptiVerse; the model writes and runs solver code (gurobi, pyomo, cvxpy, ortools and others) and a problem counts as solved only when every required variable and the objective match the ground truth within 0.1% relative error, extracted and checked by a DeepSeek-V3.2-Chat judge; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 22 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Flash | 56.86 |
| 2 | Gemini 2.5 Flash | 54.9 |
| 3 | GPT-5.2 (Thinking) | 54.9 |
| 4 | Gemini 3 Pro | 50.98 |
| 5 | DeepSeek V3.2 (Thinking) | 49.02 |
| 6 | O4 Mini | 47.06 |
| 7 | Claude Sonnet 4.5 (Thinking) | 45.1 |
| 8 | O3 | 45.1 |
| 9 | Gemini 2.5 Pro | 43.14 |
| 10 | Qwen 3 8B (Thinking) | 37.25 |
| 11 | GPT-OSS-120B | 35.29 |
| 12 | DeepSeek V3.2 (Non-reasoning) | 35.29 |
| 13 | Kimi K2 | 25.49 |
| 14 | Qwen 3 8B (Non-reasoning) | 13.73 |
| 15 | Qwen 3 Coder 30B A3B Instruct | 9.8 |
Interactive version: theaggregate.ai/benchmark?slug=optiverse-game-optimization · How It Works · Data refreshed daily, snapshot 2026-10-07.