OptiVerse - Combinatorial 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 combinatorial optimization problems; higher is better.

Metric: Accuracy (%) on the 238 combinatorial 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
1Gemini 3 Flash59.24#180
2Gemini 3 Pro57.14#146
3GPT-5.2Thinking57.14#190
4O4 Mini54.2#372
5O353.78#242
6Claude Sonnet 4.5Thinking52.52#267
7DeepSeek V3.2Thinking52.52#425
8Qwen3-235B-Thinking (OptiVerse checkpoint unspecified)51.26
9Gemini 2.5 Pro49.16#280
10Qwen 3 235B Instruct (OptiVerse checkpoint unspecified)47.06
11Qwen3-30B-Thinking (OptiVerse checkpoint unspecified)46.64
12Gemini 2.5 Flash46.22#466
13Kimi K244.54#448
14Qwen 3 8BThinking43.7#1890
15Qwen 3 30B Instruct (OptiVerse checkpoint unspecified)36.13

No result here: #2 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=optiverse-combinatorial-optimization · How It Works · Data refreshed daily, snapshot 2026-11-22.