I’ve been testing the three new models on AINFT, and honestly…
Most people are using them wrong.
They treat it like a normal chatbot.
Ask random questions → get random answers.
But once I figured out what each model is actually built for, it started making way more sense.
Here’s exactly how I use them (with real prompts I’ve tried) 🔻
𝐒𝐭𝐞𝐩 𝟏: 𝐀𝐜𝐜𝐞𝐬𝐬 𝐭𝐡𝐞 𝐌𝐨𝐝𝐞𝐥𝐬
1. Go to AINFT (chat interface)
2. Connect your wallet (TRON or BNB Chain)
3. Claim or top up credits
4. I Always Choose the Model First
Before typing anything, I ask:
“What do I actually need?”
That decides the model.
𝐇𝐨𝐰 𝐈 𝐔𝐬𝐞 𝐄𝐚𝐜𝐡 𝐌𝐨𝐝𝐞𝐥 (𝐖𝐢𝐭𝐡 𝐑𝐞𝐚𝐥 𝐏𝐫𝐨𝐦𝐩𝐭𝐬)
🔹 MiniMax-M2.5 (For structure & execution)
This is what I use when I want something I can actually act on.
Not ideas. Not theory.
Steps. Systems. Workflows.
Example prompt I’ve used:
“Create a step-by-step workflow for an AI agent that monitors ETH price, detects RSI signals, and executes trades automatically.”
What I got wasn’t fluff, it was a clear execution plan.
Now I use it for:
• automation ideas
• breaking down strategies
• turning concepts into systems
🔹 𝐊𝐢𝐦𝐢-𝐊𝟐.𝟓 (𝐅𝐨𝐫 𝐡𝐞𝐚𝐯𝐲 𝐢𝐧𝐟𝐨𝐫𝐦𝐚𝐭𝐢𝐨𝐧)
Anytime I’m dealing with too much content, I switch to this.
Docs, threads, whitepapers… it handles them easily.
Example prompt:
“Summarize this DeFi whitepaper and highlight the key risks, token utility, and revenue model.”
I’ve also used:
“Compare TRON, Ethereum, and BNB Chain in terms of fees, speed, and DeFi activity.”
It saves me hours of reading.
🔹 𝐆𝐋𝐌-𝟓 (𝐅𝐨𝐫 𝐞𝐯𝐞𝐫𝐲𝐝𝐚𝐲 𝐮𝐬𝐞)
This is my default when I just want things done quickly.
Example prompts I use:
“Write a Twitter thread explaining autonomous finance vs DeFi.”
“Generate a TRON smart contract interaction script for token swaps.”
“Explain x402 in simple terms.”
It’s fast, balanced, and reliable.
𝐒𝐭𝐞𝐩 𝟐: 𝐈 𝐂𝐨𝐦𝐛𝐢𝐧𝐞 𝐓𝐡𝐞𝐦 (𝐓𝐡𝐢𝐬 𝐈𝐬 𝐭𝐡𝐞 𝐑𝐞𝐚𝐥 𝐀𝐝𝐯𝐚𝐧𝐭𝐚𝐠𝐞)
This is where things got interesting for me.
Instead of using one model, I started doing this:
1. 𝐊𝐢𝐦𝐢 → break down information
Prompt:
“Analyze this report and extract the most important insights.”
2. 𝐌𝐢𝐧𝐢𝐌𝐚𝐱 → turn it into a system
Prompt:
“Turn these insights into a step-by-step execution strategy.”
3. 𝐆𝐋𝐌 → refine or present it
Prompt:
“Convert this into a clean, engaging Twitter thread.”
Now it feels like I’m running a workflow, not just chatting.
𝐒𝐭𝐞𝐩 𝟑: 𝐈 𝐀𝐜𝐭𝐮𝐚𝐥𝐥𝐲 𝐔𝐬𝐞 𝐭𝐡𝐞 𝐎𝐮𝐭𝐩𝐮𝐭𝐬
This is what makes AINFT different for me.
I don’t just read the answers.
I use them to:
• build ideas
• create content
• structure strategies
• plug into real workflows
𝐒𝐭𝐞𝐩 𝟒: 𝐈 𝐌𝐚𝐧𝐚𝐠𝐞 𝐂𝐫𝐞𝐝𝐢𝐭𝐬 𝐏𝐫𝐨𝐩𝐞𝐫𝐥𝐲
One mistake I made early:
Using heavy models for simple tasks.
Now I keep it simple:
• Kimi → only for deep analysis
• MiniMax → for structured thinking
• GLM → for everyday stuff
Much more efficient.
𝐑𝐞𝐚𝐥 𝐖𝐨𝐫𝐤𝐟𝐥𝐨𝐰 𝐈 𝐔𝐬𝐞
Let’s say I want to build a DeFi strategy:
• I start with Kimi
“Analyze current DeFi trends and identify profitable opportunities.”
• Then MiniMax
“Turn this into a step-by-step trading or yield strategy.”
• Then GLM
“Summarize this into a simple execution guide.”
That’s literally how I use it.
𝐅𝐢𝐧𝐚𝐥 𝐓𝐡𝐨𝐮𝐠𝐡𝐭
Once I stopped treating AINFT like a chatbot…
and started using it like a set of tools, everything changed.
You’re not just asking questions anymore.
You’re:
• analyzing
• structuring
• executing
That’s the real difference.
Official Website:
ainft.com (Enter the AI Bank section)
Product Documentation:
docs.bankofai.io/zh-Hans/