vel (@Velcrafting) posted his May "tokenmaxxing" breakdown: 4.8 billion tokens, an actual spend of $281 across five providers, and an estimate that the same usage would cost about $3,507 if it weren't subsidized by VC funding.
The stack
vel didn't share a benchmark. He shared his monthly bill, line by line.
"my tokenmaxxing for May!
4.8 Billion Tokens
Estimated Cost if not subsidized by VC funding ~ $3,507
Actual Cost spent: $281
- $200 @OpenAI -> mi amor
- $40 @grok -> like the product, not the cost
- $20 @claudeai
- $20 @MiniMax_AI
- $1 @CommandCodeAI -> Insane ROI"
Five providers, one verdict each. OpenAI got "mi amor." grok got "like the product, not the cost." Command Code, the $1 line, got "Insane ROI."
Why this matters
This is the line item that tells the story. In a $281 stack where OpenAI is $200 and grok is $40, Command Code sits at the bottom at $1 and earns the only ROI call-out in the list.
vel isn't switching providers or picking a winner. He's running everything, 4.8B tokens of it, and pricing each tool against what it returns. When the cheapest line in the stack is the one a developer tags "Insane ROI" unprompted, that's the kind of signal we can't manufacture.
Try it
1npm i -g command-codeRun cmd in your project, pick your model, and watch your usage screen. The $1/mo plan is the line item vel called Insane ROI.
Sign up for Command Code · Read the docs · See the original post
