OR-Space: leaderboard
Large language model (LLM) agents are increasingly used to assist with operations research (OR) modeling, yet existing OR-oriented benchmarks often reduce evaluation to one-shot translation from a self-contained problem statement into a mathematical formulation or solver program. Such settings abstract away two characteristics of real industrial OR workflows: persistent multi-artifact workspaces a
Metric: Build Pass@1 (self-reported). Source: benchmarklist.com. Saturation forecast: Around June 2027. 19 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 72 | #58 |
| 2 | Gemini 3 Flash | 68 | #99 |
| 3 | DeepSeek V4 Pro | 67 | #102 |
| 4 | Claude Opus 4.6 | 62 | #63 |
| 5 | GPT-5.4 | 59 | #79 |
| 6 | GPT-5 Mini | 58 | #186 |
| 7 | DeepSeek R1 0528 | 55 | #222 |
| 8 | DeepSeek V4 Flash | 54 | #137 |
| 9 | Gemini 2.5 Flash | 53 | #245 |
| 10 | Claude Sonnet 4.5 | 53 | #142 |
| 11 | GPT-5.1 | 53 | #135 |
| 12 | Claude Sonnet 4.5 (Thinking) | 53 | #142 (Claude Sonnet 4.5) |
| 13 | Qwen 3 Max | 49 | #208 |
| 14 | Gemini 2.5 Pro | 47 | #152 |
| 15 | Qwen 3 32B | 32 | #443 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=or-space · How It Works · Data refreshed daily, snapshot 2026-10-20.