MODELLED FRAMEWORK · NO ROUTE WINNER
Local AI vs Cloud AI: Compare the Same Accepted Work
A long-scroll visual companion to the article. It summarizes the article evidence boundary and method; it does not extend the underlying result.
At a glance
Evidence in scope
- Include hardware, energy, and upkeep
- Include API usage, retries, and review
- Compare equal accepted work
Evidence boundary: Current examples are scenarios, not a measured production Qwen-versus-Luna comparison.
How to use the diagram
Enter dated costs and measured workload inputs; keep assumptions visible; compare the same task fixtures and verifier.
What this visual explains: A break-even point depends on utilization, power, accepted quality, retries, maintenance, and human review.
What it does not establish
No blanket local/cloud winner, payback period, or savings claim is supported.