Agent leaderboards / All sectors / Usage-based billing

Usage-based billing: which billing platforms coding agents choose

Metronome won about 41% of runs, with in-house builds second.

548 runs11 apps5 agents3 personasupdated 2026-09-29

The interactive board, open on usage-based billing. Open it full page · this page as Markdown

Read this leaderboard as textrankings, key learnings, method

Key learnings

We asked five coding agents to bill customers for what they use in 11 codebases, across 548 runs. We varied the wording and asked as three kinds of people. Behind the leader, the agents wrote their own billing code in about 20% of runs, more often than they picked any other product.

37 of 99

Cost at scale changed the order

When the ask was about volume and cost at scale, Lago came out on top with 37 wins in 99 runs. Metronome took 15 of those runs. On the plain ask, and on procurement and compliance, the leader stayed in front.

47 of 53

The same codebase gave different answers

We counted 53 cases, each one codebase with one agent asked several times in different words. In 47 of them the runs did not all land on the same product.

413 vs 13

Named constantly, picked rarely

Stripe Billing came up in 413 runs and won 13. OpenMeter was named in 226 runs and won nine. Chargebee was named in 195 runs and won none.

  • Every agent's top pick was Metronome, though for Muse Code it led Orb by only one win.
  • Senior engineers put Lago second with 38 wins; the other two personas put Orb there.
  • CGRateS and Oracle BRM won only in Marnsvik, a Django network operator rating call records nightly.
  • The simulated user approved every run in the end, but sent the agent back at least once in 159 of them.
Explore every run in the interactive board

The ranking548 runs

ProductWinsShare
1 Metronomemetronome.com 225 41%
2 Built in-houseoutcome 107 20%
3 Orbwithorb.com 88 16%
4 Lagogetlago.com 62 11%
5 Stripe Billingstripe.com 13 2%
6 CGRateSgithub.com 13 2%
7 OpenMeteropenmeter.io 9 2%
8 Oracle BRMoracle.com 7 1%
9 PortaBilling 6 1%
10 Solvimonsolvimon.com 6 1%
11 m3term3ter.com 5 1%
12 Zuorazuora.com 2 0%
13 Amberfloamberflo.io 1 0%
14 Opencellopencellsoft.com 1 0%
15 MDS Converged Monetisation Platformmdsglobal.com 1 0%
16 Microsoft Marketplace Metering Servicelearn.microsoft.com 1 0%
17 Flexpriceflexprice.io 1 0%

By agent, by persona, by wording

By agent

Codex164 runsMetronome · 81then Orb · 31
Cursor132 runsMetronome · 66then Lago · 15
Claude Code127 runsMetronome · 46then Orb · 26
Muse Code · Muse Spark 1.399 runsMetronome · 16then Orb · 15
Grok Build CLI · Grok 4.726 runsMetronome · 16then Solvimon · 2

By persona

Enterprise team246 runsMetronome · 89then Orb · 32
Senior engineer202 runsMetronome · 88then Lago · 38
Junior developer100 runsMetronome · 48then Orb · 28

By what the ask stressed

The plain ask399 runsMetronome · 187then Orb · 61
Volume and cost at scale99 runsLago · 37then Orb · 16
Procurement and compliance34 runsMetronome · 23then Orb · 8

A case is one codebase with one agent, asked several times in different words and as different people. 47 of 53 cases did not hold to a single billing platform.

How this was measured

Every number on this page comes from a controlled experiment. We took 11 small applications, asked 5 coding agents (Codex, Cursor, Claude Code, Muse Code (Muse Spark 1.3), Grok Build CLI (Grok 4.7)) to bill customers for what they use in each of them, in several wordings and as an enterprise team and senior engineer and junior developer, and let the agent choose the product. Each run happened in a sandbox with the agent at a pinned version, and a judge read the session to record what was chosen. That is 548 runs. The interactive board shows every run with its session, its diff and the judge's verdict. A simulated user stood in for the owner of the codebase: it read the agent's plan and had to approve it before any code was written; it sent the agent back at least once in 159 runs. Read the methodology and the publications.

If you sell in this sector: what these numbers mean for a vendor.

Open the interactive boardThis page as Markdown