Vals AI

CorpFin v2

A private benchmark evaluating understanding of long-context credit agreements

Not in index
Finance
Source ↗

Models130

Top result73.2%Claude Opus 5

Top-3 spread1.6 pts

Updated12 Aug 2026

Result and cost

128 of 131 results

Results

131 results

#ModelResultCostTime

01Claude Opus 573.2% ±0.9$0.8629 s

02Claude Fable 571.8% ±0.9$1.7453 s

03Kimi K371.6% ±0.9$0.06438 s

04Muse Spark 1.171.3% ±0.9$0.10928 s

05Muse Spark 1.270.9% ±0.9$0.10830 s

06Inkling-Small69.6% ±0.9$0.02545 s

07Inkling68.6% ±0.9$0.08429 s

08Grok 4.368.5% ±0.9$0.13436 s

09GPT-5.568.4% ±0.9$0.43223 s

10Kimi K2.5 Thinking68.3% ±0.9$0.05281 s

11MiniMax M368.1% ±0.9$0.0524 s

12Grok 4.567.4% ±0.9$0.2212 min

13Claude Opus 4.6 (Adaptive)67.0% ±0.9$0.48721 s

13Claude Sonnet 567.0% ±0.9$0.51619 s

15Grok 4 Fast (Reasoning)66.9% ±0.9$0.0412 s

16Kimi K2.666.7% ±0.9$0.07380 s

16Claude Opus 4.866.7% ±0.9$0.85620 s

18Qwen 3.6 Max (preview)66.5% ±0.9$0.15647 s

19Gemini 3 Flash Preview66.4% ±0.9$0.04711 s

20Grok 4.666.2% ±0.9$0.16624 s

21GLM-5.266.1% ±0.9$0.12225 s

21Claude Opus 4.766.1% ±0.9$0.81618 s

23Grok 466.0% ±0.9$0.19130 s

23Grok 4.1 Fast (Reasoning)66.0% ±0.9$0.01628 s

25GPT-5.265.9% ±0.9$0.15426 s

25Qwen3.8 Max65.9% ±0.9$0.17947 s

27DeepSeek V4 Pro 081365.4% ±0.9$0.002233 s

28GPT-5.6 Terra65.3% ±0.9$0.1717.7 s

28Claude Sonnet 4.665.3% ±0.9$0.30718 s

28Qwen3.5 Plus Thinking65.3% ±0.9$0.0583 min

28GPT-5.465.3% ±0.9$0.23582 s

32Muse Spark65.1% ±0.9$0.01240 s

32Claude Opus 4.5 Thinking65.1% ±0.9$0.17318 s

34Gemini 3.5 Flash64.7% ±0.9$0.148.7 s

35Gemini 3.1 Pro Preview64.5% ±0.9$0.18225 s

35GLM-5.164.5% ±0.9$0.07968 s

37GPT-5.6 Sol64.4% ±0.9$0.48516 s

38GPT-5.6 Luna64.2% ±0.9$0.01917 s

39GPT-5.163.8% ±0.9$0.09235 s

40Qwen3.7 Max63.7% ±0.9$0.2220 s

40Gemini 3 Pro63.7% ±0.9$0.17920 s

40Grok 4.20 (Reasoning)63.7% ±0.9$0.1639.7 s

43Qwen3.5 Flash63.6% ±0.9$0.009635 s

44Gemini 3.6 Flash63.3% ±0.9$0.1376.3 s

45GPT-4.163.1% ±1.0$0.13210 s

46GLM-5 Thinking62.9% ±0.9$0.06869 s

47Qwen3.6-27B62.3% ±1.0$0.05463 s

48Claude Sonnet 4.5 Thinking62.0% ±1.0$0.25521 s

49Qwen3.6 Plus61.9% ±1.0$0.04736 s

49Ling 3.0 Flash61.9% ±1.0$0.006412 s

51DeepSeek V4 Flash 073161.8% ±1.0$0.01216 s

52MiMo-V2.5-Pro61.4% ±1.0$0.03719 s

52DeepSeek V4 Pro 042361.4% ±1.0$0.14979 s

54Claude Opus 4.561.3% ±1.0$0.159.4 s

55Claude Sonnet 4 Thinking61.2% ±1.0$0.2373 min

55MiniMax M2.761.2% ±1.0$0.02311 s

55GPT-5.4 nano61.2% ±1.0$0.0169.4 s

58Grok 3 Mini Fast High Reasoning61.1% ±1.0$0.01714 s

58GPT-561.1% ±1.0$0.0863 min

60Mistral Large 361.0% ±1.0$0.02449 s

60GLM-4.561.0% ±1.0$0.0382 min

62GPT-5.4 mini60.9% ±1.0$0.07652 s

63Claude Sonnet 4.560.8% ±1.0$0.23713 s

63Gemini 2.5 Pro60.8% ±1.0$0.09527 s

65Gemini 3.5 Flash-Lite60.7% ±1.0$0.0293.8 s

66Claude Haiku 4.5 Thinking60.6% ±1.0$0.08914 s

66Kimi K2 Thinking60.6% ±1.0$0.0382 min

68Claude 3.7 Sonnet Thinking60.4% ±1.0$0.23222 s

69Claude Haiku 4.560.3% ±1.0$0.0768.0 s

70GPT-5 mini60.2% ±1.0$0.02174 s

71MiMo-V2.559.9% ±1.0$0.01218 s

72Gemini 2.5 Pro Exp 03.2559.8% ±1.0$0.08616 s

72Gemini 2.5 Flash Preview Thinking59.8% ±1.0$0.02652 s

74o359.7% ±1.0$0.13319 s

74Grok 3 [Beta]59.7% ±1.0$0.1724 s

76MiniMax M2.559.6% ±1.0$0.02417 s

77Grok 3 Mini Fast Low Reasoning59.5% ±1.0$0.01711 s

78Gemini 3.1 Flash Lite Preview59.4% ±1.0$0.0234.0 s

79Gemini 2.5 Flash (Preview)59.0% ±1.0$0.02652 s

79o4-mini59.0% ±1.0$0.07418 s

81MiniMax M2.158.9% ±1.0$0.02222 s

82Mistral Medium 3.558.8% ±1.0$0.14644 s

83Grok 4 Fast Non Reasoning58.4% ±1.0$0.049.4 s

84GPT-OSS 120B58.2% ±1.0$0.009643 s

84Laguna M.158.2% ±1.0—4 min

86GPT-4.1 mini57.9% ±1.0$0.0266.4 s

87Gemini 2.5 Flash Lite Preview Thinking57.6% ±1.0$0.008654 s

88GLM-4.656.8% ±1.0$0.04578 s

89Laguna XS.256.3% ±1.0—3 min

89Gemini 2.5 Flash Lite Preview56.3% ±1.0$0.008655 s

91Qwen3 Maxqwen3-max55.9% ±1.0$0.08756 s

92DeepSeek V3 032454.7% ±1.0$0.05237 s

92Claude Sonnet 454.7% ±1.0$0.2273 min

92Trinity-Large-Thinking54.7% ±1.0$0.01912 s

95Nemotron Lightning 3p5 30B A3B54.2% ±1.0$0.004839 s

95Gemini 2.5 Flash Preview 04.1754.2% ±1.0$0.0219.0 s

97DeepSeek-R154.1% ±1.0$0.08747 s

98Claude 3.5 Sonnet53.6% ±1.0$0.2380.0 s

99GPT-OSS 20B53.1% ±1.0$0.004534 s

100Qwen3 Maxqwen3-max-preview53.0% ±1.0$0.0974 min

101Grok 4.1 Fast Non Reasoning52.5% ±1.0$0.01613 s

101DeepSeek V352.5% ±1.0$0.05228 s

103DeepSeek V3.151.5% ±1.0$0.03763 s

104Grok 251.1% ±1.0$0.11587 s

105DeepSeek V3.2 (Thinking)51.0% ±1.0$0.0382 min

106Claude 3.5 Haiku50.8% ±1.0$0.0630.0 s

107Mistral Medium50.7% ±1.0$0.02430 s

108Kimi K2 Instruct50.4% ±1.0$0.05941 s

109Llama4 Maverick Instruct Basic49.7% ±1.0$0.0144.7 s

110DeepSeek V3.247.9% ±1.0$0.0372 min

111Magistral Medium47.4% ±1.0$0.12555 s

112Llama 4 Scout 17B 16E Instruct46.8% ±1.0$0.0129.7 s

113GLM-4.746.4% ±0.9$0.03966 s

114Command A46.0% ±1.0$0.20614 s

115GPT-4o45.9% ±1.0—6.5 s

116GPT-4o mini45.5% ±1.0$0.00880.0 s

117o3-mini45.3% ±1.0$0.07831 s

118Mistral Small 3.144.2% ±1.0$0.004411 s

119Magistral Small44.0% ±1.0$0.0314 s

120Gemini 2.0 Pro Exp 02.0543.4% ±1.0$0.10519 s

121GPT-4.1 nano42.1% ±1.0$0.00655.0 s

122Jamba Large 1.641.5% ±1.0$0.1527 s

123Gemini 1.5 Pro 00240.5% ±1.0$0.13138 s

124Jamba 1.5 Large39.4% ±1.0$0.14911 s

125Meta Llama 3.1 70B Instruct Turbo38.9% ±0.9$0.0530.1 s

126Gemini 1.5 Flash 00238.2% ±1.0$0.007928 s

127Jamba Mini 1.638.0% ±1.0$0.0154.2 s

128Meta Llama 3.1 8B Instruct Turbo37.8% ±0.9$0.0110.0 s

129Jamba 1.5 Mini33.9% ±0.9$0.0152.3 s

130Gemini 2.0 Flash 00133.7% ±0.9$0.0130 s

131Gemini 1.5 Flash 00128.6% ±0.9$0.00641.2 s