How the Author Reallocated
What prompted the move was a shift in value for money: the top subscription's quota kept tightening, while the prices of rivals and APIs were falling. The new plan is a combination: keep one base subscription for daily use, route heavy jobs to pay-as-you-go APIs (pay for what you use, which actually saves money), hand some tasks to a rival model an order of magnitude cheaper, and set aside a small amount to try new tools. Total spend is unchanged, the range of capabilities covered is wider, and the dependence on a single vendor is spread out. What's good about the piece is that it carries usage data throughout, rather than being off-the-cuff impressions.
The Collective Mood of Subscription Fatigue
This made the front page because it voices many people's to-do list: stacked-up AI subscriptions are already a real expense, each vendor's quota policy keeps changing, and "playing dead and auto-renewing" is often the most expensive strategy. The solutions traded in the comments vary widely, but the methodology is consistent: first look at your real usage distribution over the past month, then match tools to task types, and recheck periodically. The AI tools market is still shifting dramatically, and spending some time to recrunch the numbers has a return no lower than tweaking prompts.
via: Hacker News