Share your thoughts, 1 month free Claude Pro on usSee more

Scientific Reasoning on GPQA Main

67.97Accuracy

LLaMA-70B

Updated 1mo ago

Evaluation Results

Method	Links
LLaMA-70B 2026.06		67.97
LLaMA-70B 2026.06		67.19
Qwen3-32B 2026.06		65.96
Qwen3-32B 2026.06		64.4
Qwen3-14B 2026.06		60.49
Qwen3-14B 2026.06		58.71
Qwen3-8B 2026.06		55.28
LLaMA-8B 2026.06		55.13
Qwen3-8B 2026.06		54.69
Qwen3-4B 2026.06		51.33
LLaMA-8B 2026.06		50.45
Qwen3-4B 2026.06		50.22
Llama 70B 2026.02		43.08
Qwen3-1.7B 2026.06		41.07
Mistral 24B 2026.02		39.95
FiMI Instruct 2026.02		37.95
Gemma 2 27B 2026.02		37.72
Qwen3-1.7B 2026.06		37.05
Official Instruct Model 2026.05		35.49
NeMo 12B IT 2026.02		34.59
Mistral 7B 2026.02		33.92
Official Instruct Model 2026.05		33.26
Min-k 2026.04		30.8
Top-nσ 2026.04		29.69
Min-k 2026.04		29.69
Opus-4.7 (xHigh) 2026.05		29.32
Min-k 2026.04		29.24
Min-p 2026.04		29.02
Top-nσ 2026.04		29.02
Min-k 2026.04		29.02
GPT-5.4 (High) 2026.05		29.02
Llama 3.1 8B 2026.02		28.57
Top-p 2026.04		28.57
Top-k 2026.04		28.35
Top-nσ 2026.04		28.35
Opus-4.7 (xHigh) 2026.05		27.83
Top-nσ 2026.04		27.46
Min-k 2026.04		27.46
GLM-4.7 & ANDES 2026.05		27.03
Top-k 2026.04		26.79
Top-nσ 2026.04		26.79
Top-nσ 2026.04		26.56
Min-k 2026.04		26.56
Opus-4.6 (1M) 2026.05		26.41
Min-p 2026.04		26.34
GPT-5.2 2026.05		25.82
GLM-4.7 & ANDES (Ours) 2026.05		25.8
Opus-4.6 2026.05		25.52
GPT-5.1-Codex-Max 2026.05		24.78
GPT-5.4 (High) 2026.05		24.33
Opus-4.6 (1M) 2026.05		22.99
Opus-4.6 2026.05		22.92
GPT-5.1-Codex-Max 2026.05		21.65
Top-p 2026.04		21.21
Min-p 2026.04		21.21
GLM-4.7 (Scaffold-only) 2026.05		20.13
GLM-4.7 (Scaffold-only) 2026.05		19.31
Min-p 2026.04		19.2
Min-k 2026.04		19.2
Min-k 2026.04		18.97
Min-k 2026.04		18.97
Min-k 2026.04		18.97
Top-k 2026.04		18.75
Top-nσ 2026.04		18.75
Min-k 2026.04		18.53
Gemini-3.1-Pro 2026.05		18.01
Top-nσ 2026.04		17.86
Top-nσ 2026.04		17.41
Top-nσ 2026.04		17.41
Top-nσ 2026.04		17.41
Min-k 2026.04		17.41
GPT-5.2 2026.05		17.41
Top-k 2026.04		17.19
Top-p 2026.04		17.19
MiniMax-M2.5 2026.05		17.19
Min-p 2026.04		16.96
Sonnet-4.5 2026.05		16.74
Gemini-3.1-Pro 2026.05		16.15
Top-p 2026.04		16.07
Top-nσ 2026.04		16.07
Top-k 2026.04		15.18
Min-p 2026.04		14.73
MiniMax-M2.1 2026.05		14.51
Base Model (Qwen3-1.7B) 2026.05		14.06
MiniMax-M2.5 2026.05		14.06
Qwen3-Max 2026.05		14.06
Kimi-K2-Thinking 2026.05		14.06
GLM-4.7 (OpenCode) 2026.05		14.06
Top-k 2026.04		12.28
MiniMax-M2.1 2026.05		12.05
Top-p 2026.04		10.71
Min-p 2026.04		10.04
Top-k 2026.04		9.6
Top-p 2026.04		7.59
Min-p 2026.04		6.92
Top-k 2026.04		5.58
Min-p 2026.04		5.58
Top-p 2026.04		5.36
Base Model (SmolLM3-3B) 2026.05		4.91
Qwen3-Max 2026.05		4.91

Showing 100 of 115 rows