The meter you didn't sign up for

You opened a tool you have paid for all year, and a small number was counting down. Credits left. Tokens remaining. Usage this cycle. You did not ask for the meter. It arrived in an update, attached to a feature you may never use, and now it sits in the corner of a product that used to just do its job.

That is the quiet shift of the last two years. Usage-based pricing started as a fairness pitch and became the default way software bills you for AI. This post is about how that happened, why it feels different from the cloud bill you signed up for, and the case for the opposite: one flat number that does not move.

#What usage-based pricing actually is

Usage-based pricing, also called consumption-based or metered billing, is a model where you pay for the exact amount of a product you consume rather than a fixed fee. The vendor picks a value metric (API calls, gigabytes, AI tokens), meters your consumption in real time, and applies a rate. You watch a dashboard to avoid a surprise bill.

For some products this is genuinely fair. Cloud infrastructure is the honest version: AWS and Google Cloud charge for compute and storage you actively chose to provision, and the bill maps cleanly to a thing you turned on. Twilio bills per SMS. An LLM API bills per token. In each case the meter measures a real, variable cost the vendor incurs on your behalf, and you opted into the variability when you adopted the tool.

The model is not the problem. The problem is where it migrated.

#How the meter ended up in your issue tracker

Usage-based billing did not stay in the infrastructure layer. As every category raced to ship AI features, the meter followed the feature in. A tool you bought as a flat seat now has an "AI" line item: a credit pool, a token allowance, a per-action charge for a copilot bolted into the sidebar. The related searches under "usage-based pricing" tell the story on their own. People are now typing "usage-based pricing Cursor" and "usage-based pricing Claude" and "usage-based pricing AI" into Google. The anxiety is specific, and it is about the tools developers use every day.

Here is the part that grates. Cloud metering is a cost you provisioned. AI metering inside a tool you already pay a seat for is a cost the vendor created by adding a feature you did not request, then handed to you to manage. You are now watching a balance drain on software whose whole job is to stay out of your way. The dashboard you check to "avoid unexpected bills" is itself the unexpected thing.

For an issue tracker, this is backwards. A tracker's job is to open fast, hold the work, and disappear. The moment it grows a meter, it stops being a system of record and becomes one more balance to babysit.

#The flat number is a position, not a discount

The honest counter to a meter is not a cheaper meter. It is no meter. Radial charges $50 per user, per year. Flat, billed annually, everything included. No tiers, no credits, no usage line, no AI surcharge, no "Contact sales." The rate you join at is the rate you keep.

To be precise, because this is the whole point: we are not claiming to be the cheapest tracker. Some tools have a lower sticker price, and a few are free. The claim is narrower and more durable. The price is one number, you can predict it a year out, and it does not climb because you used the product more or because a feature you never asked for burned through an allowance. Predictability is the feature, not the dollar amount.

The reason the bill can stay flat is structural, and it is worth saying plainly. Radial does not host, meter, or bolt in a copilot. The intelligence in your workflow is real and welcome, but it belongs to your agent, your model, and your keys, driving the tracker through a real CLI, MCP server, and API. Agents are free clients of that surface, not billable seats. There is no per-request AI cost for us to pass back to you, because we are not running AI on your behalf. No meter to install means no meter to charge.

And it is not a vibe. The stance is written down as a binding pledge: the day Radial ships a copilot, meters your usage, or charges you for AI you didn't ask for, your subscription is free.

#What "no meter" looks like in practice

The tracker is the system of record. Your agent does the work against it, for free, through a real interface. Open a terminal and have your agent file an issue the same way it would in a script or in CI:

npm i -g radial.build
radial create "Audit every tool with a usage meter" -t ENG -p high --json

That is the real CLI, returning JSON your scripts and agents can read. The same call runs over MCP at mcp.radial.build or against api.radial.build/v1 with a scoped, revocable key. Run it once or ten thousand times. The invoice is the same flat number either way, because there is nothing in that loop we meter.

#FAQ

#What is an example of usage-based pricing?

The clean examples are infrastructure: AWS and Google Cloud bill for the compute and storage you provision, Twilio bills per SMS, and an LLM API bills per token generated. In each case the meter tracks a real, variable cost you actively chose to incur. The murkier examples are the ones that arrived recently: a tool you already pay a flat seat for adding an AI credit pool or per-action charge for a copilot you did not request. Same mechanism, very different fairness.

#What is usage-based billing pricing?

It is billing where you pay only for what you consume in each cycle, also called metered or pay-as-you-go pricing. The vendor picks a value metric, meters your usage in real time, and charges a per-unit rate. It is the same model behind your electricity bill or an Uber fare. The question worth asking of any software vendor is whether the metered thing is a cost you chose to incur, or one the vendor added and handed back to you to watch.

#How does usage-based pricing work?

The vendor selects a measurable metric (API calls, gigabytes, tokens, or copilot actions), tracks your consumption continuously, and multiplies usage by a predefined rate to produce the bill. You typically monitor a dashboard to avoid a surprise charge. The flip side, the one the marketing skips, is that the meter makes your cost variable and slightly unpredictable, which is exactly the thing a flat annual price removes.

#What are the four types of pricing?

The common buckets are flat-rate (one fixed price), tiered (a few fixed plans), per-seat (priced by user), and usage-based (priced by consumption). Most modern SaaS blends them, and the AI era pushed many tools toward a hybrid: a per-seat base plus a usage meter for AI. Radial is deliberately the simplest version, flat per-user, billed annually, with no usage component to track.

#Is usage-based pricing bad?

No. For infrastructure where cost genuinely scales with what you turn on, it is the fair model. It becomes a problem when it is grafted onto a tool you bought as a flat seat, to bill for an AI feature you did not ask for. The objection here is not to metering in general. It is to the meter showing up uninvited in the tools that are supposed to stay out of your way.

#One number, the same every January

If you are tired of checking a balance on software whose entire job is to disappear, the answer is not a tool with a cheaper meter. It is a tool with no meter. One flat price, locked, with your agents riding free through real interfaces and a pledge that turns the promise into a commitment.

See the one number on pricing, or read why a tracker is boring on purpose and built to look the same next year.