Alignment on ArenaHard (pass@1)
95.7pass@1Gemini-2.5 Flash-Thinking
Evaluation Results
| Method | Links | |
|---|---|---|
| Gemini-2.5 Flash-ThinkingThinking Mode=true2025.12 | 95.7 | |
| DeepSeek-R1 0528 671BParameters=671B, Thinking Mode=true2025.12 | 95.1 | |
| Qwen3 14BParameters=14B2025.12 | 91.7 | |
| Nemotron-Cascade 14B-ThinkingParameters=14B, Thinking Mode=true2025.12 | 89.5 | |
| Nemotron Cascade-8BParameters=8B, Thinking Mode=true2025.12 | 87.9 | |
| Qwen3 8BParameters=8B2025.12 | 85.8 | |
| Nemotron-Nano 9B-v2Parameters=9B-v22025.12 | 74.6 |