| # | Model | Provider | Type | Score | |
|---|---|---|---|---|---|
| 1 | Claude Mythos 5 | Anthropic | Closed | claude-mythos-5 | |
| 2 | Kimi K3 | Moonshot AI | Closed | kimi-3 | |
| 3 | Claude Opus 4.7 (Adaptive) | Anthropic | Closed | claude-opus-4-7-max | |
| 4 | Claude Opus 4.8 | Anthropic · Closed | 80.5% | ||
| 5 | Inkling | Thinking Machines Lab · Open weight | 78.1% | ||
| 6 | Claude Sonnet 5 | Anthropic · Closed | 77% |