Graduate-Level Google-Proof Q&A

A challenging dataset of 448 multiple-choice questions written by domain experts in biology, physics, and chemistry. Designed to be difficult even for skilled non-experts with access to Google.

In index
Knowledge
Best in knowledge

Models141

Top result96.0%GPT-6 Astra

Top-3 spread0.5 pts

Year—

Result and price

77 of 141 results

Results

141 results

#ModelResult

01GPT-6 Astra96.0%

02Sakana Fugu-Ultra95.5%

02Sakana Fugu95.5%

04GPT-5.6 Sol94.6%

05Kimi K3.1 Pro94.2%

05Claude Opus 4.7 (Adaptive)94.2%

07Claude Mythos 594.1%

08GPT-5.593.6%

08Claude Opus 4.893.6%

10Kimi K393.5%

11Inkling 2 Small93.0%

12GPT-5.6 Terra92.9%

13GPT-5.492.8%

13Ornith-1.5-397B92.8%

15Qwen3.8 Max92.6%

16Qwen3.7 Max92.4%

16GPT-5.292.4%

18Hy4 preview92.3%

18GPT-5.6 Luna92.3%

20Gemini 3.5 Flash92.2%

21Qwen3.8-Flash-Next91.7%

22Claude Opus 4.691.3%

23GLM-5.291.2%

24Qwen3.8-Omni-Flash91.0%

25DeepSeek V4.1 Flash 101790.9%

26Kimi K2.690.5%

27Qwen3.6 Plus90.4%

28Qwen3.7 Plus90.3%

29DeepSeek V4 Pro 081390.1%

29Grok 4.390.1%

31Claude Sonnet 4.689.9%

31Interfaze Beta89.9%

33Inkling-Small89.5%

34Qwen3.8-27B89.2%

34Ornith-1.5-35B-A3B89.2%

36Qwen3.5 397B88.4%

37DeepSeek V4 Flash 073188.1%

38GPT-5.4 mini88.0%

39Inkling87.9%

40Qwen3.6-27B87.8%

41Kimi K2.5 (Reasoning)87.6%

41Kimi K2.587.6%

43Hy3 Preview87.2%

44Claude Opus 4.587.0%

44Nemotron 3 Ultra87.0%

46Qwen3.5-122B-A10B86.6%

47Ornith-1.5-9B86.4%

48GLM-586.0%

48Qwen3.6-35B-A3B86.0%

50Ternary Bonsai 2 27B85.8%

51GLM-4.785.7%

52Qwen3.5-27B85.5%

53Ling 3.0 Flash85.0%

54Gemma 4 31B84.3%

55MAI-Thinking-184.2%

55Qwen3.5-35B-A3B84.2%

57Ling 3.0 Flash FP884.0%

58MiMo-V2-Flash83.7%

59Claude Sonnet 4.583.4%

60Gemini 2.5 Pro83.0%

61GPT-5.4 nano82.8%

62Gemma 4 26B A4B82.3%

63Qwen3 235B 2507 (Reasoning)81.1%

64DeepSeek-R1-052881.0%

65Grok 3 Mini80.3%

66GPT-OSS 120B80.1%

66DeepSeek V3.1 (Reasoning)80.1%

68Claude 4 Opus Thinking79.6%

69Nemotron 3 Super 100B79.2%

70o1-pro79.0%

71Gemma 4 12B78.8%

72Claude 3.7 Sonnet Thinking78.2%

73o4-mini77.6%

74Qwen3 235B 250777.5%

75o3-mini77.2%

76Nemotron Ultra 253B76.0%

77o175.7%

78Nemotron 3.5 Lightning 30B A3B NVFP475.6%

79Grok 3 [Beta]75.4%

80GLM-4.7-Flash75.2%

81Kimi K275.1%

82DeepSeek V3.174.9%

82Claude 4 Opus74.9%

84LFM2.5-24B-A2B74.4%

85Nemotron 3 Nano 30B73.0%

86Nemotron 3 Nano Omni 30B A3B72.2%

87GPT-OSS 20B71.5%

87DeepSeek-R171.5%

89Ministral 3 14B (Reasoning)71.2%

90Qwen3 235B-A22B (Reasoning)71.1%

91ZAYA1-8B71.0%

92Magistral Medium70.8%

93Spark-X2.6-4B69.9%

94GPT-4.569.5%

95Phi-4-reasoning-plus68.9%

96Qwen3 32B (Reasoning)68.4%

97Magistral Small68.2%

98Claude 3.7 Sonnet68.0%

99Granite 4.2 30B66.4%

100GPT-4.166.3%

101Phi-4-reasoning65.8%

102QwQ-32B65.6%

103Gemini 2.0 Flash65.2%

104Gemini 2.5 Flash-Lite64.6%

105GPT-4.1 mini64.2%

106Granite 4.2 8B64.1%

107LFM3-1.7B63.4%

108Qwen3 235B-A22B62.9%

109Qwen2.5-Max60.1%

110o1-mini60.0%

111Claude 3.5 Sonnet59.4%

112DeepSeek V359.1%

113Ling 2.6 Flash59.0%

114Gemma 4 E4B58.6%

115Mellum2-12B-A2.5B-Thinking57.6%

116ZAYA1-74B-Preview57.3%

117Phi-456.1%

118Grok 256.0%

119Granite 4.2 3B54.8%

120Qwen3 32B54.6%

121Grok 2 mini51.0%

122Llama 3.3 70B50.5%

122Gemini 2.0 Flash-Lite50.5%

124GPT-4.1 nano50.3%

125Gemini 1.5 Flash50.0%

126Llama 3.1 70B46.7%

127Qwen2.5-VL-32B46.0%

127Mistral Small 3.146.0%

129Gemma 4 E2B43.4%

129Soofi S 30B-A3B43.4%

131Claude 3.5 Haiku41.6%

132Mellum2-12B-A2.5B-Instruct40.9%

133LFM2.5-350M30.6%

134Llama 3.1 8B30.4%

134Phi-3.5-mini30.4%

136Granite-4.0-1B29.9%

137Granite-4.0-H-1B29.7%

138Granite-4.0-H-350M26.1%

139LFM2.5-VL-450M25.7%

140LFM2.5-230M25.4%

141Granite-4.0-350M24.1%