OptiVerse - Dynamic Optimization: leaderboard

OptiVerse (2026) is a benchmark of textbook-sourced optimization problems that models solve by writing and running solver code, a problem counting as solved only when every required value matches the ground truth within 0.1%. This board is accuracy in percent on the dynamic optimization problems; higher is better.

Metric: Accuracy (%) on the 146 dynamic 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 February 2027. 18 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1GPT-5.2Thinking57.53#185
2Gemini 3 Pro56.85#145
3O356.85#237
4Gemini 3 Flash55.48#175
5O4 Mini55.48#368
6Qwen3-235B-Thinking (OptiVerse checkpoint unspecified)52.74
7DeepSeek V3.2Thinking49.32#419
8Gemini 2.5 Flash48.63#463
9Gemini 2.5 Pro47.95#276
10Claude Sonnet 4.5Thinking47.95#261
11Qwen3-30B-Thinking (OptiVerse checkpoint unspecified)46.58
12Qwen 3 235B Instruct (OptiVerse checkpoint unspecified)41.1
13Kimi K239.73#443
14Qwen 3 8BThinking39.73#1873
15Qwen 3 30B Instruct (OptiVerse checkpoint unspecified)34.25

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=optiverse-dynamic-optimization · How It Works · Data refreshed daily, snapshot 2026-11-21.