Design Optimization on 10 optimization shared-core domains
67.5Success RateOpus 4.6
Evaluation Results
| Method | Links | |
|---|---|---|
| Opus 4.6Turns=20 (Opt.)2026.03 | 67.5 | |
| Sonnet 4.6Turns=20 (Opt.)2026.03 | 63 | |
| Sonnet 4.5Turns=20 (Opt.)2026.03 | 61.5 | |
| GPT-5.2Turns=20 (Opt.)2026.03 | 58.5 | |
| Gemini 3.1 ProTurns=20 (Opt.)2026.03 | 56 | |
| Gemini 2.0 FlashTurns=20 (Opt.)2026.03 | 48 | |
| GPT-4oTurns=20 (Opt.)2026.03 | 41.5 | |
| Opus 4.6Turns=1 (Opt.)2026.03 | 34.5 | |
| Gemini 3.1 ProTurns=1 (Opt.)2026.03 | 31.3 | |
| Sonnet 4.6Turns=1 (Opt.)2026.03 | 31.2 | |
| Sonnet 4.5Turns=1 (Opt.)2026.03 | 30.5 | |
| GPT-5.2Turns=1 (Opt.)2026.03 | 29 | |
| Gemini 2.0 FlashTurns=1 (Opt.)2026.03 | 22.2 | |
| GPT-4oTurns=1 (Opt.)2026.03 | 17.8 |