Knowledge-intensive Reasoning on SuperGPQA
72.7Overall ScoreGemini-3-Flash
Evaluation Results
| Method | Links | |
|---|---|---|
| Gemini-3-FlashTier=High2026.06 | 72.7 | |
| Seed2.0 Lite2026.06 | 67.5 | |
| Qwen3.5Parameter Count=35BA3B2026.05 | 63.4 | |
| Gemma4Parameter Count=26BA4B2026.05 | 61.88 | |
| Seed2.0 Mini2026.06 | 61.6 | |
| GPT-5-miniTier=High2026.06 | 60.5 | |
| SenseNova-U1Parameter Count=30BA3B, Configuration Variant=Think2026.05 | 59.71 | |
| Qwen3.5Parameter Count=9B2026.05 | 58.2 | |
| Qwen3VLParameter Count=30BA3B, Configuration Variant=Think2026.05 | 56.4 | |
| Qwen3VLParameter Count=8B, Configuration Variant=Think2026.05 | 51.2 | |
| SenseNova-U1Parameter Count=8B, Configuration Variant=Think2026.05 | 49.67 | |
| SCOPE-RLBase Model=Qwen3-4B2025.10 | 38.6 | |
| GRPOBase Model=Qwen3-4B2025.10 | 38.4 | |
| Entropy-RegBase Model=Qwen3-4B2025.10 | 38.2 | |
| Entropy-AdvBase Model=Qwen3-4B2025.10 | 37.8 | |
| KL-covBase Model=Qwen3-4B2025.10 | 37.6 | |
| DAPOBase Model=Qwen3-4B2025.10 | 37 | |
| CISPOBase Model=Qwen3-4B2025.10 | 36.8 | |
| BaseBase Model=Qwen3-4B2025.10 | 32.8 | |
| SCOPE-RLBase Model=Qwen2.5-7B2025.10 | 27.2 | |
| GRPOBase Model=Qwen2.5-7B2025.10 | 27 | |
| SCOPE-RLBase Model=Qwen2.5-Math-7B2025.10 | 26.8 | |
| Entropy-RegBase Model=Qwen2.5-7B2025.10 | 26.8 | |
| KL-covBase Model=Qwen2.5-Math-7B2025.10 | 26.6 | |
| Entropy-AdvBase Model=Qwen2.5-7B2025.10 | 26.6 | |
| Entropy-RegBase Model=Qwen2.5-Math-7B2025.10 | 26.2 | |
| Entropy-AdvBase Model=Qwen2.5-Math-7B2025.10 | 26.2 | |
| CISPOBase Model=Qwen2.5-Math-7B2025.10 | 26.2 | |
| DAPOBase Model=Qwen2.5-Math-7B2025.10 | 26 | |
| KL-covBase Model=Qwen2.5-7B2025.10 | 26 | |
| GRPOBase Model=Qwen2.5-Math-7B2025.10 | 25.8 | |
| CISPOBase Model=Qwen2.5-7B2025.10 | 25.8 | |
| DAPOBase Model=Qwen2.5-7B2025.10 | 24.8 | |
| BaseBase Model=Qwen2.5-Math-7B2025.10 | 19.4 | |
| BaseBase Model=Qwen2.5-7B2025.10 | 18.6 |