ReactionOptimisation
Task 35 / 47
reizman_suzuki_pareto
Pareto optimization on the Reizman Suzuki emulator over catalyst choice and continuous conditions to improve conflicting chemical metrics. Multi-objective black-box search under chemistry-side constraints is scored through the benchmark's evaluator, mirroring co-design of recipe and operating point.
Model leaderboard
Updated v1 results · 2026-09-15 · raw score, higher is better.
| # | Participant | Raw score | Medal |
|---|---|---|---|
| 1 | GLM-5 | 82.9901 | Gold |
| 2 | Claude Opus 4.6 | 82.3427 | Silver |
| 3 | GPT-5.4 | 82.246123 | Bronze |
| 4 | DeepSeek V3.2 | 82.0329 | — |
| 5 | Qwen3 Coder Next | 81.4666 | — |
| 6 | Seed 2.0 Pro | 79.7011 | — |
| 7 | Gemini 3.1 Pro Preview | 79.473 | — |
| 8 | Grok 4.20 | 63.5202 | — |
Framework results · original paper
Historical results from the original evaluators · normalized score (0–100).
| # | Participant | Score |
|---|---|---|
| 1 | GPT-OSS + OpenEvolve | 100.0 |
| 2 | Claude Opus 4.6 + OpenEvolve | 44.0 |
| 3 | Claude Opus 4.6 + ABMCTS | 43.8 |
| 4 | Claude Opus 4.6 + ShinkaiEvolve | 36.8 |
| 5 | GPT-OSS + ShinkaiEvolve | 20.0 |
| 6 | GPT-OSS + ABMCTS | 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.