Benchmarks / Professional

Finance Agent v2

Evaluating agents on core financial analyst tasks using the FAB v2 harness

77models
10%Open Index weight
2026-10-12last updated
Vals AIpublisher

Finance Agent v2

Higher is better · best effort per model

Chart settings
77 of 77 selected

full table below

#ModelEffortScoreCost / taskHarness
1Gemini 4high65.4%$4.38–
2Gemini 3.8 Flashhigh61.4%$2.00–
3Muse Spark 1.2xhigh60.6%$0.770–
4Muse Spark 1.3max60.0%$0.757–
5Wonder Fullmax59.1%$2.39–
6Gemini 3.7 Flashhigh59.0%$1.48–
7Fable 5.1max58.9%$8.35–
8Opus 5max58.6%$5.12–
9Opus 5.5max58.6%$9.22–
10Sonnet 5.5max58.1%$6.55–
11Gemini 3.5 Flashhigh57.9%$2.51–
12GLM 5.3 Flashmax57.9%$0.048–
13MiMo V2.6 Prohigh57.3%$0.199–
14Muse Spark 1.1xhigh57.2%$0.723–
15Wondermax56.6%$2.67–
16Fable 5max56.3%$8.06–
17Gemini 3.6 Flashhigh56.3%$1.40–
18MiMo V2.6 Flashhigh56.3%$0.072–
19GLM 5.3max55.8%$1.07–
20Hy4–55.1%$0.590–
21GPT-5.6 Lunamax55.0%$0.279–
22Ling 3.0 Flash Fin–54.9%$0.045–
23Mistral Large 4high54.7%$1.20–
24GPT-5.6 Terramax54.4%$3.63–
25Haiku 5.5max54.1%$1.64–
26Opus 4.8max53.9%$4.22–
27Sonnet 5max53.9%$0.751–
28GPT-5.6 Solmax53.8%$1.25–
29Grok 4.6high53.7%$1.66–
30GPT-6 Astramax53.5%$6.82–
31DeepSeek V4.1 Flashhigh53.5%$0.207–
32Kimi K3max53.1%$1.91–
33Grok 4.7xhigh52.3%$2.72–
34GPT-6.1 Solmax52.0%$1.62–
35Ember-1–51.9%$1.95–
36GPT-5.5xhigh51.8%$4.15–
37Opus 4.7high51.5%$4.03–
38Sonnet 4.6max51.0%$2.41–
39Step 5 Preview–50.7%$0.619–
40Qwen 3.8 Maxmax50.6%$1.24–
41DeepSeek V4 Pro 0813max50.4%$0.880–
42GPT-6 Lunamax49.9%$0.124–
43GLM 5.2max49.7%$0.714–
44DeepSeek V4 Flash 0731high49.5%$0.276–
45GPT-6 Solmax49.0%$2.12–
46Qwen 3.8 27Bxhigh48.6%$0.749–
47Grok 4.5high48.3%$1.16–
48MiniMax M3max48.3%$0.323–
49Qwen 3.7 Maxmax47.8%$0.553–
50Gemini 3.5 Flash-Litehigh47.4%$0.390–
51Inklingmax46.6%$0.996–
52GPT-5.4 Minixhigh45.4%$1.20–
53Kimi K2.6max44.9%$1.14–
54GLM 5.1max44.8%$0.618–
55DeepSeek V4max44.1%$0.879–
56Gemini 3.1 Prohigh43.0%$1.59–
57Gemini 3 Flashhigh42.6%$0.571–
58MiMo V2.5 Proxhigh41.5%$0.206–
59Inkling Smallxhigh41.3%$0.230–
60Qwen 3.6 Plusmax40.8%$0.814–
61Qwen 3.7 Plushigh38.2%$0.358–
62GPT-5.4 Nanohigh38.2%$0.159–
63Grok 4.3high37.7%$0.848–
64Nemotron 3 Ultra 550B A55Bmax37.7%––
65MiMo V2.5xhigh36.7%$0.087–
66Kimi K2.5max35.8%$0.294–
67Mistral Medium 3.5high32.1%$1.91–
68Haiku 4.5xhigh31.0%$0.602–
69Ling 3.0 Flash–30.3%$0.050–
70Gemini 3.1 Flash-Litehigh30.0%$0.142–
71Grok 4.20xhigh28.5%$0.792–
72MiniMax M2.7high27.9%$0.183–
73Laguna M.1–25.0%$0.818–
74Mercury 2.5high18.9%$0.106–
75Nemotron 3.5 Lightning–18.5%$0.029–
76Laguna XS.2–15.6%$0.192–
77Command Anone9.0%$1.63–