Benchmarks / Professional

Excel Modeling Benchmark

Evaluating agents on Excel-based financial modeling tasks

73models
–Open Index weight
2026-10-12last updated
Vals AIpublisher

Excel Modeling Benchmark

Higher is better · best effort per model

Chart settings
73 of 73 selected

full table below

#ModelEffortScoreCost / taskHarness
1Fable 5.1max76.7%$15.93–
2Opus 5.5max75.9%$11.02–
3Sonnet 5.5max75.7%$8.79–
4Gemini 4high75.2%$11.80–
5Fable 5max73.7%$12.35–
6Opus 5max73.6%$6.07–
7GPT-5.6 Solmax72.3%$6.01–
8Gemini 3.8 Flashhigh72.2%$8.24–
9Wonder Fullmax71.9%$4.58–
10GPT-6 Astramax71.7%$5.83–
11GPT-6 Solmax71.5%$1.28–
12Gemini 3.7 Flashhigh71.3%$5.69–
13GPT-6.1 Solmax70.8%$1.14–
14Opus 4.8max69.4%$12.06–
15GPT-6 Lunamax68.5%$0.142–
16Muse Spark 1.3max67.4%$2.55–
17GPT-5.6 Lunamax67.1%$0.367–
18Wondermax67.1%$5.60–
19Grok 4.7xhigh67.0%$6.48–
20Kimi K3max66.7%$3.07–
21Sonnet 5max66.3%$10.29–
22GPT-5.6 Terramax66.2%$2.19–
23MiMo V2.6 Flashhigh65.5%$0.109–
24Gemini 3.6 Flashhigh65.4%$3.89–
25GPT-5.5xhigh64.5%$3.27–
26Opus 4.7max63.7%$12.76–
27Gemini 3.5 Flashhigh63.5%$5.17–
28Ember-1–63.4%$2.46–
29MiMo V2.6 Prohigh62.9%$0.329–
30Grok 4.6high62.7%$3.06–
31GLM 5.2max61.5%$4.47–
32Sonnet 4.6max60.2%$7.23–
33Qwen 3.8 Maxmax60.1%$2.66–
34Qwen 3.8 27Bxhigh59.7%$4.56–
35Haiku 5.5max59.0%$1.15–
36Hy4–58.1%$0.787–
37Kimi K2.6max57.9%$2.20–
38DeepSeek V4.1 Flashhigh57.2%$0.234–
39Step 5 Preview–57.1%$1.33–
40Qwen 3.7 Maxmax57.0%$2.86–
41Muse Spark 1.2xhigh57.0%$2.31–
42DeepSeek V4 Flash 0731high57.0%$0.384–
43Muse Spark 1.1xhigh56.4%$1.11–
44GLM 5.3max56.3%$3.79–
45Mistral Large 4high56.2%$4.84–
46GLM 5.3 Flashmax55.9%$0.713–
47MiMo V2.5 Proxhigh55.2%$0.221–
48MiMo V2.5xhigh55.1%$0.083–
49Grok 4.5high52.9%$1.55–
50DeepSeek V4 Pro 0813max52.8%$1.08–
51Gemini 3.1 Prohigh52.6%$3.77–
52DeepSeek V4max51.6%$0.797–
53Qwen 3.7 Plushigh49.3%$0.598–
54MiniMax M3max47.8%$2.09–
55GPT-5.4 Minixhigh45.4%$2.15–
56GPT-5.4 Nanohigh44.8%$0.573–
57Gemini 3.5 Flash-Litehigh43.2%$1.23–
58Inklingmax38.4%$1.67–
59Ling 3.0 Flash Fin–36.8%$0.180–
60Qwen 3.6 Plusmax32.9%$2.87–
61Inkling Smallxhigh31.8%$0.249–
62Nemotron 3 Ultra 550B A55Bmax31.7%––
63Kimi K2.5max28.5%$0.787–
64MiniMax M2.7high28.5%$1.66–
65Gemini 3 Flashhigh26.0%$0.453–
66Ling 3.0 Flash–25.0%$0.180–
67Haiku 4.5xhigh22.7%$1.03–
68Grok 4.3high18.4%$0.634–
69Nemotron 3.5 Lightning–17.0%$0.182–
70Grok 4.20xhigh11.7%$0.470–
71Mistral Medium 3.5high11.2%$10.60–
72Mercury 2.5high8.8%$0.418–
73Gemini 3.1 Flash-Litehigh8.6%$0.069–