最佳数学 AI 模型

168 个模型 · 数学指数。至少两项数学结果。方法

上榜模型

2026年11月

#模型数学指数输出 / 1M速度

01Claude Fable 5.279.385.6$40.0064/s

02GPT-6.1 Sol78.278.0$10.0034/s

03GPT-6 Astra78.078.0$50.0034/s

04Claude Opus 5.577.778.6$20.0076/s

05Gemini 4 Carbon77.282.7$10.0031/s

06Claude Fable 5.177.076.2$50.0069/s

07GPT-6.1 Terra76.977.1$5.0046/s

08GPT-5.5 Pro76.374.8$180.0019/s

09Claude Fable 576.274.0$50.0056/s

10Claude Sonnet 5.576.177.2$10.00107/s

11Gemini 4 Argon75.978.5$10.0028/s

12GPT-6 Sol75.872.8$10.0061/s

13GPT-5.4 Pro75.373.3$180.0021/s

14GPT-5.6 Sol75.173.0$22.0064/s

15Claude Opus 574.873.8$25.0048/s

16GPT-5.574.769.1$30.0052/s

17Muse Wonder73.573.1$20.0080/s

18GPT-5.6 Terra72.868.9$12.0066/s

19Claude Opus 4.871.267.7$25.0069/s

20Muse Spark 1.368.670.1$4.2576/s

21GPT-6 Luna68.565.5$0.5066/s

22GPT-5.6 Luna68.465.5$1.2051/s

23Claude Haiku 5.568.366.3$0.50159/s

24GPT-5.468.064.8$15.0038/s

25Qwen3.8 Max67.665.6—41/s

26Gemini 3.7 Flash66.868.5$3.7585/s

27Kimi K366.666.9$16.0044/s

28DeepSeek V4.1 Flash 101765.468.5$0.40261/s

29Grok 4.665.367.8$6.0056/s

30Muse Spark Open64.965.1$4.2586/s

31Qwen3.8 Max (0902)64.863.4$6.0033/s

31Claude Sonnet 564.864.2$10.0075/s

33Muse Spark 1.264.464.9$4.25123/s

34Gemini 3.8 Flash64.367.8$4.1368/s

34Claude Opus 4.764.362.0$25.0047/s

36DeepSeek V4 Pro 081364.165.5$4.4442/s

37Claude Opus 4.663.858.0$25.0036/s

38GLM-5.363.165.8$1.4086/s

39GPT-5.262.659.0$14.0045/s

40Qwen3.7 Max62.160.5$4.4368/s

41Grok 4.761.869.1$6.0051/s

41Grok 4.561.863.7$6.0053/s

43Gemini 3.1 Pro61.762.3$12.00151/s

44Mistral Large 461.660.7$2.10113/s

45DeepSeek V4 Flash 073161.561.9$0.3061/s

45Gemini 3.5 Flash61.564.2$9.00137/s

47Gemini 3.6 Flash60.763.7$3.75121/s

48Gemini 3.1 Pro Preview60.461.5$12.0077/s

49Muse Spark60.259.8——

50Kimi K2.660.058.2$3.4181/s

51GLM-5.259.660.6$6.80111/s

52Gemini 3 Pro58.257.9$12.00109/s

53DeepSeek V4 Pro 042357.657.6$1.7889/s

54GPT-5.157.555.3$10.0050/s

55Grok 4.2057.452.8$2.5072/s

56GLM-5.3-Flash57.363.1$0.078122/s

56Gemini 3 Flash57.355.0$3.00198/s

58Kimi K2.7 Code57.057.7$3.5035/s

59Solar Open 256.753.3——

60Qwen3.8-27B56.359.2$1.88113/s

61GPT-5.4 nano55.950.9$1.2591/s

62Qwen3.7 Plus55.457.7$1.2851/s

63GPT-555.355.0$10.0076/s

63Inkling-Small55.355.1$1.2080/s

65GLM-5.154.956.0$4.40102/s

65Qwen3.6 Plus54.954.6$1.9544/s

67Ling 3.0 Flash54.550.9$0.06373/s

68Inkling54.154.8$4.0585/s

68Claude Sonnet 4.654.157.0$15.0044/s

70GPT-5.4 mini53.954.5$4.5095/s

71Kimi K2.553.852.2$3.3236/s

72Qwen3.5 397B53.653.8$3.6059/s

73Qwen 3.6 Max (preview)53.257.2$6.1675/s

74Ripple-65B Preview52.752.3$0.6063/s

75MAI-Thinking-152.653.5——

76GLM-552.551.6$3.2045/s

77GPT-5 mini52.248.4$2.0078/s

77Kimi K2.5 (Reasoning)52.253.0$3.00—

79GPT-OSS 120B51.543.4$0.6034/s

80Claude Opus 4.551.352.0$25.0039/s

81Ternary Bonsai 2 27B51.152.2$0.5021/s

82DeepSeek V3.2 (Thinking)51.049.4$2.19—

82Qwen3.6-27B51.052.1$3.2060/s

82K-EXAONE 2.051.048.8——

85Grok 450.550.0—78/s

85Kimi K2 Thinking50.548.2$2.5048/s

85Spark-X2.6-4B50.548.7——

85o4-mini50.544.7$4.40122/s

85Gemini 3.1 Flash-Lite50.551.9$1.50134/s

90GLM-4.749.150.0$2.2028/s

90o349.151.8$8.0065/s

92Qwen3.6-35B-A3B49.050.6$0.9065/s

93GLM-4.548.445.7$2.2025/s

94MiniMax M2.747.751.2$0.8435/s

95Granite 4.2 30B47.541.6——

96DeepSeek V3.246.844.2$0.3819/s

97Sarvam 105B46.636.7—191/s

97Claude Sonnet 4.546.647.9$15.0045/s

99DeepSeek V3.1 (Reasoning)46.543.0——

99Gemini 3.5 Flash-Lite46.552.0$2.50130/s

101Qwen3 Max46.345.0$3.9062/s

102o3-mini46.140.0$4.40173/s

103Claude Sonnet 4 Thinking46.046.9——

103Gemini 2.5 Pro46.046.6$10.0091/s

105GPT-OSS 20B45.941.7$0.13310/s

105DeepSeek-R1-052845.944.6$2.1520/s

107GPT-5 nano45.841.9$0.4081/s

108GLM-4.644.941.8$2.0030/s

109Gemini 2.5 Flash44.135.0$2.5095/s

110Claude 4 Opus Thinking43.947.0——

110MiniCPM5-2B43.937.1——

110o143.944.5$60.0097/s

110Grok 3 Mini43.944.3$0.5065/s

110MiniMax M2.143.949.1$1.2034/s

115DeepSeek-R143.838.9$2.5018/s

116ZAYA1-8B43.738.2——

117Kimi K2 Instruct43.343.2——

118Nemotron Ultra 253B42.535.9——

119Claude Opus 4.142.047.5$75.0012/s

120Granite 4.2 8B41.937.6$0.15101/s

120QwQ-32B41.936.3——

122Magistral Medium41.833.7——

123MiniMax M341.655.7$1.2079/s

124Magistral Small41.331.4——

125Claude Haiku 4.540.543.2$5.0088/s

126Claude Opus 439.443.9——

127Claude Sonnet 438.943.0$15.0049/s

128Grok 3 [Beta]38.239.3——

129DeepSeek V3 032438.037.1——

130Sarvam 30B37.930.3——

131o1-mini37.233.5$4.40—

132Claude 3.7 Sonnet Thinking36.641.4——

133GPT-4.1 mini36.534.5$1.6072/s

134Kimi K236.039.6$2.3037/s

135GPT-4.135.938.0$8.0074/s

136Phi-4-reasoning35.527.1—41/s

137Granite 4.2 3B34.333.2——

138GPT-4.534.038.8$150.00—

139LongCat-Flash-Lite-Sparse32.539.3——

140Mistral Medium 332.227.1$2.0094/s

140Qwen3 235B-A22B32.238.1$1.8244/s

142Llama4 Maverick Instruct Basic30.336.0——

143Claude 3.7 Sonnet30.037.7——

144DeepSeek V3.129.937.9—98/s

145GPT-4.1 nano29.126.3$0.4056/s

146Llama 4 Maverick28.634.4$1.0061/s

147Gemma 3 27B28.525.3—31/s

148Gemini 2.0 Flash28.129.5——

149DeepSeek V327.229.9$1.1037/s

150LFM2.5-8B-A1B27.130.6—335/s

151Claude 3.5 Sonnet25.232.2$15.00—

152Grok 225.029.1——

153Llama 4 Scout24.226.8$0.3063/s

154Phi-423.722.4$0.1493/s

155GPT-4o23.428.7$10.0066/s

156Command A23.023.9——

157GPT-4o mini22.026.2$0.6056/s

157Mistral Large 222.027.4$6.0070/s

159MiniCPM5-1B21.915.5——

160Llama 3.1 405B18.927.0—29/s

161Gemini 1.5 Flash18.020.2——

162Mistral Small 3.116.724.2—28/s

163Claude 3.5 Haiku16.224.1——

164Llama 3.3 70B15.628.2——

165Claude 3 Opus14.423.3$75.00—

166Llama 3.1 70B13.924.8——

167Llama 3.1 8B10.017.9——

168Claude 3 Haiku8.122.0$1.2573/s