Last month I checked my statements — I spent 820 yuan (~$115) on AI subs, more than I spent on food delivery
I stared at the bill and asked myself: which model is actually doing the work, and which just has a big name? Then a friend tossed me a link, and I realized this was never something I needed to agonize over.
It's a free price-comparison site that auto-updates every day
The site is called bestmodelforyourbudget (bestmodelforyourbudget.terrydjony.com), built by Terry Djony. It scrapes the latest prices and capability scores from every major AI model daily, then tells you straight up based on your budget: "who's actually worth buying this month."
A Xiaohongshu blogger I know, Lin Xia, used to manually compare GPT-4o and Claude output quality every week. Then she bookmarked the link and glanced at it before each copywriting session — within a month, her subscription spend dropped 40% with no quality loss.
Cost to start today: ¥0, 5 minutes, zero technical barrier
Money: Completely free, no paywall.
Time: 5 minutes to figure it out the first time. I got stuck here too — a wall of scores and no clue which one actually matters. Don't worry, the author explains every single score; it clicks after one look.
Technical barrier: Zero. If you can open a webpage and type a number, you're good.
First step: Open bestmodelforyourbudget.terrydjony.com, type in how much you're willing to spend on AI per month (say, 200), hit enter.
Advice by stage
If you're just starting out with no stable clients yet: this tool isn't your top priority right now. Just use the free tiers (ChatGPT, Claude, and Gemini all have free plans) — that's plenty. Come back to compare prices when you're spending over 100/month; skipping it for now is totally fine.
If you already have 1-2 stable clients and spend 200-1000/month on AI: use it now. I'd bookmark it and check on the 1st of every month — model prices shift fast, and you should switch to whoever's cheapest the moment it happens.
If you're scaling up and the team is starting to split AI costs: screenshot the recommendation results and drop them in the team group chat. Let the copywriter, scriptwriter, and designer each pick what fits them — no need for everyone to use the same model.