OptiVerse - Hard: 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 hard problems; higher is better.

Metric: Accuracy (%) on the 300 hard-level 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 December 2027. 20 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1Gemini 3 Pro27#145
2Gemini 3 Flash25.33#175
3GPT-5.2Thinking25.33#185
4Claude Sonnet 4.5Thinking21.67#261
5DeepSeek V3.2Thinking21.33#419
6Qwen3-235B-Thinking (OptiVerse checkpoint unspecified)21.33
7O320.67#237
8O4 Mini20.67#368
9Qwen3-30B-Thinking (OptiVerse checkpoint unspecified)20.33
10Gemini 2.5 Pro20#276
11Gemini 2.5 Flash18.67#463
12GPT-OSS-120B18.67#704
13Qwen 3 235B Instruct (OptiVerse checkpoint unspecified)16.67
14Kimi K216.33#443
15Qwen 3 8BThinking16#1873

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

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