JOR-Bench - MAMO Complex LP: leaderboard
JOR-Bench (2026) translates the MAMO Complex LP linear programming problems into Japanese with LLM translation and human post-editing. Models write an optimization program that is executed with OR-Tools available, and a problem counts as solved when the printed value matches the reference; accuracy in percent, higher is better.
Metric: Accuracy (%; MAMO Complex LP linear programming problems, Japanese translation with human post-editing; the model writes a standalone Python program in a fenced block, run with OR-Tools available (300-second and 32 GB limits), correct when the printed value matches the reference within 1e-6; higher is better). Source: arxiv.org. Saturation forecast: Around February 2027. 7 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-OSS-120B | 73.5 | #700 |
| 2 | Qwen 3 32B | 50 | #960 |
| 3 | Qwen3-Swallow-32B-RL-v0.2 | 37.5 | #900 |
| 4 | Qwen 3 8B | 34.5 | #1830 |
| 5 | Qwen3-Swallow-8B-RL-v0.2 | 10.5 | #1517 |
| 6 | NVIDIA-Nemotron-Nano-9B-v2-Japanese | 1 | #1486 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=jor-bench-mamo-complex-lp · How It Works · Data refreshed daily, snapshot 2026-11-13.