Science Question Answering on GPQA Diamond (test)
81.5AccuracyDC-RS + Janus
Evaluation results
| Method | Links | |
|---|---|---|
| DC-RS + JanusBackbone LLM=Qwen3-8B2026.06 | 81.5 | |
| ExpeL + JanusBackbone LLM=Qwen3-8B2026.06 | 78.5 | |
| o1Model Type=API only, Fine-tuning examples=N.A.2025.01 | 77.3 | |
| ExpeL + JanusBackbone LLM=DeepSeek-V4-Flash2026.06 | 76.4 | |
| ExpRAGBackbone LLM=DeepSeek-V4-Flash2026.06 | 74.9 | |
| DC-RSBackbone LLM=Qwen3-8B2026.06 | 73.4 | |
| o1-previewModel Type=API only, Fine-tuning examples=N.A.2025.01 | 73.3 | |
| DC-RS + JanusBackbone LLM=DeepSeek-V4-Flash2026.06 | 73.3 | |
| ExpeLBackbone LLM=Qwen3-8B2026.06 | 72.9 | |
| ExpeLBackbone LLM=DeepSeek-V4-Flash2026.06 | 71.8 | |
| DeepSeek r1Model Type=Open Weights, Fine-tuning examples=>800K2025.01 | 71.5 | |
| Memory-freeBackbone LLM=DeepSeek-V4-Flash2026.06 | 71.2 | |
| DC-RSBackbone LLM=DeepSeek-V4-Flash2026.06 | 70.7 | |
| ExpRAGBackbone LLM=Qwen3-8B2026.06 | 69.2 | |
| r1-distillModel Type=Open Weights, Fine-tuning examples=800K2025.01 | 62.1 | |
| o1-miniModel Type=API only, Fine-tuning examples=N.A.2025.01 | 60 | |
| s1-32BModel Type=Open Weights and Open Data, Fine-tuning examples=1K, Budget Forcing=true2025.01 | 59.6 | |
| Bespoke-32BModel Type=Open Weights and Open Data, Fine-tuning examples=17K2025.01 | 58.1 | |
| Memory-freeBackbone LLM=Qwen3-8B2026.06 | 57.4 | |
| Sky-T1Model Type=Open Weights and Open Data, Fine-tuning examples=17K2025.01 | 56.8 | |
| s1 w/o BFModel Type=Open Weights and Open Data, Fine-tuning examples=1K, Budget Forcing=false2025.01 | 56.6 | |
| QwQ-32BModel Type=Open Weights, Fine-tuning examples=N.A.2025.01 | 54.5 | |
| Qwen2.5-32B-InstructModel Type=Open Weights, Fine-tuning examples=N.A.2025.01 | 49 |