Vals AI

TaxEval v2

A Vals-created set of questions and responses to tax questions

Not in index
Finance
Source ↗

Models142

Top result80.4%Muse Spark 1.2

Top-3 spread2.7 pts

Updated1 Sep 2026

Result and cost

140 of 143 results

Results

143 results

#ModelResultCostTime

01Muse Spark 1.280.4% ±0.8$0.01448 s

02Muse Spark 1.179.7% ±0.8$0.01533 s

03Muse Spark77.7% ±0.8$0.001958 s

04Claude Sonnet 4.677.1% ±0.8$0.1282 min

05Claude Fable 576.9% ±0.8$0.2257 s

06GPT-5.6 Terra76.2% ±0.8$0.03637 s

06GPT-5.6 Luna76.2% ±0.8$0.01277 s

08Claude Fable 5.176.0% ±0.8$0.34786 s

08Claude Opus 4.6 (Adaptive)76.0% ±0.8$0.07179 s

10Grok 3 [Beta]75.9% ±0.8$0.01713 s

11GPT-5.275.8% ±0.8$0.06660 s

12Kimi K375.7% ±0.8$0.0782 min

12Grok 4 Fast (Reasoning)75.7% ±0.8$0.001310 s

14Claude Opus 4.875.6% ±0.8$0.15970 s

14Claude Sonnet 575.6% ±0.8$0.3023 min

14GLM-5.3-Flash75.6% ±0.8$0.001553 s

14Qwen3.8 Max75.6% ±0.8$0.0764 min

18Inkling-Small75.5% ±0.8$0.0184 min

19Qwen3.7 Max75.3% ±0.8$0.02858 s

19Inkling75.3% ±0.8$0.1033 min

19Claude Opus 4.775.3% ±0.8$0.12562 s

22GPT-5 mini75.2% ±0.8$0.01237 s

23Claude Opus 575.1% ±0.8$0.11458 s

23GPT-4.175.1% ±0.8$0.00627.9 s

25GPT-5.575.0% ±0.9$0.062 min

26Gemini 3.6 Flash74.9% ±0.8$0.0212 s

26GPT-5.174.9% ±0.9$0.007244 s

26Claude Opus 4.5 Thinking74.9% ±0.8$0.21947 s

29o4-mini74.8% ±0.9$0.007115 s

29GPT-5.6 Sol74.8% ±0.9$0.13862 s

31Gemini 3.7 Flash74.7% ±0.8$0.0197.8 s

31Qwen3.6 Plus74.7% ±0.9$0.01171 s

31Kimi K2.674.7% ±0.8$0.0294 min

34o374.6% ±0.9$0.01324 s

35GPT-4o74.5% ±0.9$0.00585.9 s

36Gemini 3.8 Flash74.4% ±0.8$0.0248.8 s

36Gemini 3.5 Flash74.4% ±0.8$0.03820 s

38Claude Opus 4.574.3% ±0.9$0.05915 s

38o174.3% ±0.9$0.12219 s

40Kimi K2.5 Thinking74.2% ±0.8$0.0142 min

41Grok 4.20 (Reasoning)74.1% ±0.9$0.01713 s

42Claude 3.7 Sonnet Thinking74.0% ±0.9$0.05143 s

42GPT-5.474.0% ±0.9$0.15779 s

42Qwen3 Maxqwen3-max-preview74.0% ±0.9$0.006721 s

45Gemini 3 Flash Preview73.9% ±0.9$0.005812 s

46MiMo-V2.5-Pro73.8% ±0.9$0.002563 s

47Claude Opus 4.1 Thinking73.7% ±0.9$0.09327 s

48Qwen3 Maxqwen3-max73.5% ±0.9$0.006529 s

49GPT-573.4% ±0.9$0.05479 s

50GLM-5.273.3% ±0.9$0.01565 s

50Claude Sonnet 4.5 Thinking73.3% ±0.9$0.04148 s

52Grok 4.1 Fast (Reasoning)73.1% ±0.9$0.001331 s

52DeepSeek V4 Pro 081373.1% ±0.9$0.0464 min

52Mistral Large 373.1% ±0.9$0.001630 s

55Grok 3 Mini Fast Low Reasoning73.0% ±0.9$0.001112 s

56Gemini 2.5 Pro Exp 03.2572.9% ±0.9$0.01122 s

56Gemini 3.1 Pro Preview72.9% ±0.9$0.03142 s

58MiniMax M372.7% ±0.9$0.00732 min

58Gemini 2.5 Flash (Preview)72.7% ±0.9$0.00316.3 s

60Gemini 3.5 Flash-Lite72.6% ±0.9$0.00738.8 s

60Gemini 3 Pro72.6% ±0.9$0.03424 s

62GLM-4.572.4% ±0.9$0.00962 min

62Claude 3.7 Sonnet72.4% ±0.9$0.00646.3 s

62Gemini 2.5 Flash Preview Thinking72.4% ±0.9$0.003111 s

62GLM-5.372.4% ±0.9$0.0322 min

66DeepSeek-R172.3% ±0.9$0.0123 min

67Qwen3.5 Flash72.2% ±0.9$0.003173 s

68DeepSeek V4 Pro 042372.1% ±0.9$0.0333 min

69Claude Sonnet 4 Thinking72.0% ±0.9$0.02932 s

70Claude Opus 471.9% ±0.9$0.03715 s

70GPT-4.1 mini71.9% ±0.9$0.00125.9 s

72MiMo-V2.571.8% ±0.9$0.0007326 s

72Gemini 3.1 Flash Lite Preview71.8% ±0.9$0.000967.6 s

74Kimi K2 Thinking71.7% ±0.9$0.009161 s

74Grok 4.571.7% ±0.9$0.0193 min

76GPT-OSS 120B71.6% ±0.9$0.00348 s

76Grok 4 Fast Non Reasoning71.6% ±0.8$0.000574.3 s

78Claude Opus 4.171.5% ±0.9$0.0433 s

79Qwen3.6-27B71.3% ±0.9$0.032 min

80GPT-5.4 mini71.2% ±0.9$0.006845 s

80GLM-5.171.2% ±0.9$0.01784 s

80Gemini 2.5 Flash Preview 04.1771.2% ±0.9$0.00265.4 s

83Grok 3 Mini Fast High Reasoning71.1% ±0.9$0.002125 s

83Grok 4.671.1% ±0.9$0.01748 s

83DeepSeek V3 032471.1% ±0.9$0.0006534 s

86Qwen3.8-27B70.8% ±0.9$0.0462 min

86Grok 4.370.8% ±0.9$0.0147 min

88DeepSeek V4 Flash 073170.7% ±0.9$0.009872 s

89Ling 3.0 Flash70.6% ±0.9$0.000667.5 s

89Qwen3 235B-A22B70.6% ±0.9$0.00462 min

91Gemini 2.5 Flash Preview 04.17 Thinking70.5% ±0.9$0.007515 s

92Mistral Medium70.3% ±0.9$0.001411 s

93Kimi K2 Instruct70.2% ±0.9$0.001513 s

93Claude 3.5 Sonnet70.2% ±0.9$0.0065.9 s

95GLM-5 Thinking70.0% ±0.9$0.0222 min

96Gemini 2.0 Flash Thinking Exp 01.2169.8% ±0.9$0.0008313 s

97Claude Sonnet 469.6% ±0.9$0.007612 s

98o3-mini69.4% ±0.9$0.01874 s

99GLM-4.768.8% ±0.9$0.0143 min

100MiniMax M2.568.2% ±0.9$0.008673 s

100DeepSeek V3.2 (Thinking)68.2% ±0.9$0.00242 min

100Gemini 2.0 Pro Exp 02.0568.2% ±0.9$0.0049.0 s

103Mistral Medium 3.568.0% ±0.9$0.0792 min

104DeepSeek V367.9% ±0.9$0.0005628 s

105Gemini 2.0 Flash Exp67.7% ±0.9$0.00037.5 s

106Claude Haiku 4.5 Thinking67.5% ±0.9$0.02748 s

107GPT-5.4 nano67.4% ±0.9$0.0006416 s

107GPT-5 nano67.4% ±0.9$0.004970 s

109Grok 267.0% ±0.9$0.006911 s

110MiniMax M2.766.6% ±0.9$0.003529 s

110Llama4 Maverick Instruct Basic66.6% ±0.9$0.0005325 s

112MiniMax M2.166.4% ±0.9$0.00912 min

113GLM-4.666.2% ±0.8$0.009364 s

113Gemini 2.5 Flash Lite Preview66.2% ±0.9$0.000483.6 s

115Gemini 2.0 Flash 00165.3% ±0.9$0.00035.7 s

116Grok 465.1% ±1.0$0.0783 min

117Gemini 2.5 Flash Lite Preview Thinking64.7% ±0.9$0.000513.9 s

118Mistral Large 241163.8% ±0.9$0.004112 s

119Llama 3.3 Nemotron Super 49B V1 42e84561 Thinking63.7% ±0.9—33 s

119GPT-OSS 20B63.7% ±0.9$0.002349 s

121Magistral Medium61.9% ±1.0$0.02546 s

121Grok 4.1 Fast Non Reasoning61.9% ±0.9$0.000379.3 s

123Command A61.4% ±1.0$0.006511 s

124Jamba Large 1.660.9% ±0.9$0.005216 s

124Meta Llama 3.1 405B Instruct Turbo60.9% ±1.0$0.001924 s

126GPT-4.1 nano60.8% ±1.0$0.000293.0 s

127GPT-4o mini60.5% ±1.0$0.000328.9 s

128Magistral Small60.3% ±1.0$0.003916 s

129Llama 3.3 Nemotron Super 49B V1 42e8456160.2% ±1.0—18 s

130Gemini 1.5 Pro 00259.5% ±1.0$0.00259.9 s

131Llama 3.3 70B Instruct Turbo59.4% ±1.0$0.000533.8 s

132Laguna XS.259.0% ±1.0$0.0003127 s

133Mistral Small 3.158.3% ±1.0$0.000188.0 s

134Jamba 1.5 Large58.2% ±1.0$0.005420 s

135Claude 3.5 Haiku57.4% ±1.0$0.00135.0 s

136Meta Llama 3.1 70B Instruct Turbo56.2% ±1.0$0.000494.3 s

137Llama 4 Scout 17B 16E Instruct55.2% ±1.0$0.000387.0 s

138Mistral Small 240249.1% ±1.0$0.000369.0 s

139Gemini 1.5 Flash 00248.2% ±1.0$0.000163.7 s

140Jamba Mini 1.644.6% ±1.0$0.000264.3 s

141Jamba 1.5 Mini41.9% ±1.0$0.000284.8 s

142Meta Llama 3.1 8B Instruct Turbo32.3% ±0.9$0.000112.5 s

143Laguna M.11.6% ±0.3—4 min