benchgap
Agentic · tools

Finance Agent v2 leaderboard

As of 2026-10-13, the highest measured score on Finance Agent v2 is 65.4% by Gemini 4 Argon. 22 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#ModelScoreSource
1Gemini 4 Argon65.4%measured
2Claude Sonnet 5.562.4%estimated ± 4.5 pp, low confidence
3Claude Opus 5.561.9%estimated ± 4.5 pp, low confidence
4DeepSeek V4.1 Flash61.8%estimated ± 4.5 pp, low confidence
5GPT-6 Astra61.7%estimated ± 4.5 pp, low confidence
6Gemini 3.8 Flash61.4%measured
7Grok 4.661.4%estimated ± 4.5 pp, low confidence
8Grok 4.761.2%estimated ± 4.5 pp, low confidence
9GPT-6.1 Sol61.1%estimated ± 4.5 pp, low confidence
10GLM-5.360.5%estimated ± 4.5 pp, low confidence
11GPT-6 Sol60.4%estimated ± 4.5 pp, low confidence
12GLM-5.3-Flash60.2%estimated ± 4.5 pp, low confidence
13Claude Fable 5.160.0%estimated ± 4.5 pp, low confidence
14MiMo-V2.6-Pro59.9%estimated ± 4.5 pp, low confidence
15Ling 3.0 Flash Fin59.8%measured
16Kimi K359.8%estimated ± 4.5 pp, low confidence
17Muse Spark 1.359.7%estimated ± 4.5 pp, low confidence
18Muse Wonder Full59.1%measured
19GPT-6 Luna58.8%estimated ± 4.5 pp, low confidence
20Step 5 Preview58.4%estimated ± 4.5 pp, low confidence
21Qwen3.8-27B57.9%estimated ± 4.5 pp, low confidence
22Ling 3.1 Flash57.9%measured
23Gemini 3.5 Flash57.9%measured
24Muse Spark 1.157.2%measured
25Muse Wonder56.6%measured
26Claude Haiku 5.555.5%estimated ± 4.5 pp, low confidence
27Mistral Large 454.7%measured
28Claude Opus 4.853.9%measured
29MiniMax M352.8%estimated ± 4.5 pp, low confidence
30Muse Glimmer 30B50.0%estimated ± 4.5 pp, low confidence
31Inkling49.7%estimated ± 4.5 pp, low confidence
32Nemotron 3 Ultra49.3%estimated ± 4.5 pp, low confidence

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General