Minds in Finance.

Mint-Cu
Mint-Ag
98.33%
RFC-Bench
89.04%
FinSearchComp T2
54.07%
9B + 27B

01
02
03
04

98.33%
RFC-Bench · Mint-Ag

54.07%
FinSearchComp T3 · Mint-Ag

69.86%
FinSearchComp T2 · Mint-Cu 9B

Mint-Agent benchmark results
Mint-Ag Mint-Cu Brand-colored baselines

BizFinBench

Mint-Ag55.71
Mint-Cu53.86
GPT-5.6-Sol51.57
Gemini-3.550.71
Claude-Opus49.71
Qwen-3.748.43
Cursor-Grok48.29

FinanceBench

Mint-Ag91.33
Claude-Opus90.00
Mint-Cu90.00
DSV4-Flash88.67
GLM-5.288.00
Qwen-3.788.00
Cursor-Grok88.00

RFC-Bench

Mint-Ag98.33
Mint-Cu96.67
Claude-Opus95.33
GPT-5.6-Sol94.67
GLM-5.294.33
Qwen-3.793.33
Kimi-K2.792.67

FinanceAgentBench v1.1

Mint-Ag76.00
DSV4-Flash72.00
Cursor 2.572.00
MiniMax-M370.00
Mint-Cu68.00
Claude-Opus66.00
GPT-5.6-Sol66.00

FinanceAgentBench v2

Mint-Ag60.49
GPT-5.6-Sol56.79
Claude-Opus55.56
Codex GPT-5.654.32
Cursor-Grok53.09
GLM-5.251.85
Cursor 2.549.38

FinSearchComp T2

Mint-Ag89.04
Cursor-Grok81.74
Codex GPT-5.680.37
Cursor 2.577.63
Tongyi-DR72.15
Mint-Cu69.86
GLM-5.268.49

Mint-Agent trinity tech stack
Tech Stack · 01

Atomic financial task construction

Long-horizon agentic task construction and verification
Tech Stack · 02

MintHarness stateful execution loop

Tech Stack · 03

Dual-expert training and unification pipeline

CASE 01

AMD quarterly revenue guidance trajectory

CASE 02

TSMC Q1 2025 revenue projection trajectory

mintfinance@tutamail.com