Transparent AI Pricing: Credits, Not Marked-Up Keys

Most AI products pick one of two billing models, and they feel very different on the invoice:

• Metered model bills. You wire up a provider account, the tool runs against it, and your cost rises and falls with usage you can't easily predict — often with a reseller margin baked into each call.
• Included credits. AI usage is part of the plan. You know what a month costs before it starts, and the AI just works without a separate account to manage.

PulseCraft is the second kind. The model runs underneath; you spend credits, not keys. The difference shows up most on a busy month: metered billing punishes the weeks you produce the most, exactly when the work is paying off; credits keep the number flat.

• Predictable cost. Your plan tells you what the month costs. Credits don't surprise you with a spike because one busy week generated more posts.
• Nothing to provision. No provider sign-up, no key rotation, no quota juggling. The engine has what it needs the moment you start.
• One bill, plainly stated. AI usage is on your PulseCraft invoice as credits — not buried in a separate dashboard you have to reconcile.

For a team generating real volume, predictability is worth more than chasing a per-token rate you have to babysit. Finance teams in particular prefer a number they can forecast over a usage bill that arrives after the spend is already made.

The quiet issue with metered, bring-a-key-the-vendor-marks-up models is that the price you pay per call isn't the price the vendor pays — there's often a margin folded into each token. That margin is invisible by design: you see "AI usage" on the bill, not the spread between wholesale and retail.

Included credits sidestep this. The cost is set when you pick a plan, so there's no per-call spread to reverse-engineer. You're paying for a known allotment, not an opaque multiplier on usage you can't audit.

You'll see tools advertise "free AI captions" or a generous-sounding free AI allowance. It's worth reading the fine print, because "free" usually means one of three things:

• A capped trial of a single feature — free for the simplest task (a caption), with the real work (multi-platform generation, images) gated behind a paid tier or metered usage.
• A loss-leader on top of a metered base — the AI is free, but you still wire up and pay for a provider account underneath, so the "free" applies only to the tool's thin wrapper.
• A volume cliff — generous until you hit a threshold, then it converts to per-call billing exactly when you start producing real volume.

None of these is dishonest on its own, but they make the real cost of producing at volume hard to see up front. Included credits are deliberately the opposite: the number is set when you choose a plan, and there's no cliff hiding behind the word "free."

Whatever tool you're evaluating — ours included — these questions cut through the marketing to the real cost structure:

1. Is AI usage included in the plan price, or billed separately?
2. If separate, is it metered per call, and is there a markup over the underlying provider's rate?
3. Do I need to provision and manage my own provider account to use it?
4. What happens on a high-volume month — does my bill scale, and by how much?
5. Is the heavy work (multi-platform generation, image generation) included, or only the cheap stuff (a single caption)?
6. Can I bring my own keys if I want control or already have negotiated rates?
7. Can I see usage before and after a run, so cost isn't a surprise?

A vendor with transparent pricing will answer all seven plainly. If any answer is evasive — especially on markup and the high-volume case — that's the signal to dig deeper before you commit.

The practical question isn't "is credit pricing good" in the abstract — it's "which plan matches how much I actually produce." A simple way to think about it:

• Light producers (a few posts a week, one or two networks) rarely approach their allotment; the smallest paid tier is usually plenty, and the value is the predictability, not the ceiling.
• Steady teams (daily posting across several networks) want enough headroom that a busy week doesn't pause the engine — pick the tier whose allotment covers your peak week, not your average.
• High-volume agencies (many clients, many networks, frequent image generation) consume the most per source; here the right move is the tier built for that throughput, plus the option to top up rather than re-plan mid-month.

Because the cost is fixed at the plan and usage is visible in your workspace, you can start conservative and move up once you can see your real consumption — there's no penalty for guessing low first. That's the opposite of metered billing, where guessing wrong shows up as a surprise on the invoice instead of a planned upgrade.

One honesty note on scope: credits cover the AI work the engine does — analyzing sources, writing, adapting, formatting, and generating images. They are not a metered platform fee and they don't pay for anything outside the engine, like the networks' own API limits or any advertising spend you choose to run. Keeping that boundary clear is part of the point: you should always be able to say exactly what a credit is for, which is hard to do when AI cost is folded opaquely into a per-token line on someone else's bill.