JobShop
Task 15 / 47
abz
Classical job-shop scheduling on the ABZ benchmark family (Adams, Balas, Zawack 1988): sequence operations on machines to minimize makespan or tardiness—strongly NP-hard combinatorial optimization. Scoring compares against published bounds/optima for relative gap, a core manufacturing scheduling challenge.
Model leaderboard
Updated v1 results · 2026-09-15 · raw score, higher is better.
| # | Participant | Raw score | Medal |
|---|---|---|---|
| 1 | Claude Opus 4.6 | 96.1035 | Gold |
| 2 | GPT-5.4 | 91.231431 | Silver |
| 3 | GLM-5 | 88.4924 | Bronze |
| 4 | DeepSeek V3.2 | 88.3614 | — |
| 5 | Grok 4.20 | 87.6717 | — |
| 6 | Gemini 3.1 Pro Preview | 86.751 | — |
| 7 | Seed 2.0 Pro | 86.672 | — |
| 8 | Qwen3 Coder Next | 85.603 | — |
Framework results · original paper
Historical results from the original evaluators · normalized score (0–100).
| # | Participant | Score |
|---|---|---|
| 1 | Claude Opus 4.6 + ABMCTS | 100.0 |
| 2 | Claude Opus 4.6 + ShinkaiEvolve | 90.0 |
| 3 | Claude Opus 4.6 + OpenEvolve | 83.2 |
| 4 | GPT-OSS + ShinkaiEvolve | 28.0 |
| 5 | GPT-OSS + ABMCTS | 14.9 |
| 6 | GPT-OSS + OpenEvolve | 0.0 |
Model results use the updated scores and medal thresholds. No valid score earns zero medal credit. Historical framework scores use the original evaluators and have not been updated alongside the model results.