| # | Model | Provider | Type | Score | |
|---|---|---|---|---|---|
| 1 | Kimi K2.5 (Reasoning) | Moonshot AI | Closed | 96.1% | |
| 2 | Kimi K2.5 | Moonshot AI | Open | 96.1% | |
| 3 | GLM-4.7 | Z.AI | Open | 95.7% | |
| 4 | MiMo-V2-Flash | Xiaomi | Open | 94.1% | |
| 5 | Claude Sonnet 4.5 | Anthropic | Closed | 87% | |
| 6 | Exaone 4.0 32B | LG AI Research | Open | 85.3% | |
| 7 | Nemotron 3 Nano Omni 30B A3B | NVIDIA | Open | 82.1% | |
| 8 | LFM2.5-8B-A1B | LiquidAI | Open | 42.5% | |
| 9 | MiniCPM5-1B | OpenBMB | Open | 40.4% |