Question Answering on MMLU (Accuracy)
92AccuracyGPT-4.1
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-4.12025.10 | 92 | |
| Claude Sonnet 42025.10 | 89.5 | |
| DOCTOR-R12025.10 | 85 | |
| Baichuan-M2-32B2025.10 | 84 | |
| Evo 8BShots=5, E2E Latency (s)=8.6, Inference Speed (tokens/s)=522026.02 | 76.8 | |
| Qwen2.5 7BShots=5, E2E Latency (s)=8.1, Inference Speed (tokens/s)=462026.02 | 74.5 | |
| OriginBase Model=Qwen-3-8B2026.02 | 73.01 | |
| FULLModel=Qwen3-8b, Zero-shot protocol=true2025.12 | 72.97 | |
| DoctorAgent-RL2025.10 | 72.5 | |
| Qwen3-8Bbase_model=true2025.10 | 70 | |
| BD3-LM 7BShots=5, E2E Latency (s)=14.2, Inference Speed (tokens/s)=282026.02 | 70 | |
| KUDABase Model=Qwen-3-8B2026.02 | 68.92 | |
| SPONModel=Qwen3-8b, Zero-shot protocol=true2025.12 | 68.79 | |
| TEALModel=Qwen3-8b, Zero-shot protocol=true2025.12 | 68.78 | |
| FULLModel=Llama3-8b, Zero-shot protocol=true2025.12 | 68.05 | |
| UltraMedical-8B2025.10 | 67.5 | |
| LLaDA 8BShots=5, E2E Latency (s)=21.8, Inference Speed (tokens/s)=162026.02 | 65.6 | |
| LLaMA3 8BShots=5, E2E Latency (s)=7.4, Inference Speed (tokens/s)=582026.02 | 65.1 | |
| MDLM 7BShots=5, E2E Latency (s)=18.9, Inference Speed (tokens/s)=222026.02 | 64 | |
| OriginBase Model=Llama-3.1-8B2026.02 | 63.31 | |
| SPONModel=Llama3-8b, Zero-shot protocol=true2025.12 | 60.95 | |
| KUDABase Model=Llama-3.1-8B2026.02 | 60.64 | |
| Med42-v2-8B2025.10 | 60.5 | |
| RMUBase Model=Qwen-3-8B2026.02 | 60.02 | |
| TEALModel=Llama3-8b, Zero-shot protocol=true2025.12 | 59.69 | |
| FULLModel=Mistral-v2-7B, Zero-shot protocol=true2025.12 | 59.05 | |
| BaseModel=Mistral-7B, Zero-shot=true2026.03 | 59 | |
| BaseModel=Zephyr-7B-β, Zero-shot=true2026.03 | 58.5 | |
| RMUBase Model=Llama-3.1-8B2026.02 | 58.33 | |
| OriginBase Model=Zephyr-7B-Beta2026.02 | 58.13 | |
| RMUModel=Mistral-7B, Zero-shot=true2026.03 | 57.7 | |
| ASUModel=Zephyr-7B-β, Zero-shot=true2026.03 | 57.5 | |
| ASUModel=Mistral-7B, Zero-shot=true2026.03 | 57.2 | |
| KUDABase Model=Zephyr-7B-Beta2026.02 | 57.14 | |
| RMUModel=Zephyr-7B-β, Zero-shot=true2026.03 | 57 | |
| RMUBase Model=Zephyr-7B-Beta2026.02 | 56.9 | |
| SPONModel=Mistral-v2-7B, Zero-shot protocol=true2025.12 | 55.78 | |
| TEALModel=Mistral-v2-7B, Zero-shot protocol=true2025.12 | 55.37 | |
| SCRUBModel=Zephyr-7B-β, Zero-shot=true2026.03 | 51.2 | |
| SimNPOBase Model=Zephyr-7B-Beta2026.02 | 48.26 | |
| ARD 7BShots=5, E2E Latency (s)=32.5, Inference Speed (tokens/s)=122026.02 | 46.2 | |
| NPOBase Model=Zephyr-7B-Beta2026.02 | 45.71 | |
| LLMUModel=Zephyr-7B-β, Zero-shot=true2026.03 | 44.7 | |
| SSDModel=Zephyr-7B-β, Zero-shot=true2026.03 | 40.7 | |
| GABase Model=Zephyr-7B-Beta2026.02 | 33.02 | |
| Gemini-2.5-Flash2025.10 | 20.5 |