$0.25 Per Review: Why We Chose Usage-Based Pricing

Most developer tools charge per seat. $24 per user per month. $49 per user per month. $59 per user per month. The logic is straightforward: each developer who uses the tool gets a license, and the company pays for each license.

This model made sense when developer tools were things developers actively used — IDEs, debuggers, profilers. Tools that occupied screen time and required human interaction. Per-seat pricing tracked per-human value.

AI agents break this model. An AI code reviewer does not occupy screen time. It runs asynchronously, processes pull requests in the background, and posts results. A developer does not "use" the tool in the traditional sense — the tool uses itself. Charging per seat for a tool that runs autonomously makes as little sense as charging per seat for a CI system.

The per-seat problem

Per-seat pricing for AI tools creates three specific problems.

It penalizes adoption. Adding a new team member to your AI code review tool costs $24-59 per month, regardless of whether that team member opens one PR or fifty. This creates friction at exactly the moment you want zero friction — when a team is deciding whether to adopt the tool more broadly.

It misaligns incentives.The vendor is incentivized to maximize seat count. The customer is incentivized to minimize it. This produces the worst possible outcome: tools that are rolled out to some team members but not others, creating inconsistent workflows and reducing the tool's effectiveness.

It does not reflect value. An AI code reviewer delivers value per review, not per seat. A team of 5 that opens 200 PRs per month gets more value than a team of 20 that opens 50 PRs per month. Per-seat pricing charges the second team four times more.

Usage-based pricing for AI agents

Ellipsis charges $0.25 per code review. No per-seat fees. No platform fees. Unlimited team members, unlimited repositories. You pay for the work the agent does, not for the number of humans on your team.

This aligns every incentive correctly:

• Adding team members is free. Roll the tool out to your entire org without budget negotiations.
• Cost scales with value. If the agent reviews more PRs, you are getting more value, and you pay proportionally more. If it reviews fewer, you pay less.
• No shelfware. You never pay for capacity you do not use. There is no incentive to undercount seats or restrict access.

The economics

Let's run the numbers for a typical team.

A 10-person engineering team opens roughly 80-120 pull requests per month. At $0.25 per review, that is $20-30 per month for automated code review on every PR. The same team using a per-seat tool at $29 per user per month would pay $290.

For a 50-person team opening 400 PRs per month: $100 with Ellipsis vs. $1,450+ with per-seat tools.

# Monthly cost comparison
#
# Team size: 10 engineers, ~100 PRs/month
#
# Per-seat tool A ($49/user):     $490/mo
# Per-seat tool B ($29/user):     $290/mo
# Per-seat tool C ($24/user):     $240/mo
#
# Ellipsis ($0.25/review):         $25/mo
#
# Annual savings vs cheapest:    $2,580

The cost difference is not marginal. It is an order of magnitude. This matters for startups watching every dollar and for enterprises deploying to hundreds of engineers.

Why not free?

Running AI agents costs real money. Each code review involves cloning a repository, running a language model with significant context, and posting structured results. Our cost per review is roughly $0.08-0.12, depending on the size of the PR and the complexity of the analysis.

Charging $0.25 per review gives us a sustainable margin while keeping the price low enough that cost is never a blocker for adoption. We could charge less, but we would rather build a sustainable business than race to the bottom on price.

What about other agents?

Code review is our most popular agent, but Ellipsis runs six agent types: code review, bug fixes, code generation, Q&A, reports, and analytics. All agents are included at no additional cost. The $0.25-per-review pricing covers the code review agent; other agents are included as part of the platform.

As we add more agent types, they will be included in the same pricing model. The platform scales with usage, not with feature count.

Spending controls

Predictability matters. We offer monthly spending caps configurable in the dashboard. Set a cap at $50 per month, and Ellipsis stops running reviews once the cap is reached. No surprises.

For enterprises with procurement requirements, we offer annual contracts with committed spend. Contact team@ellipsis.dev for details.

Pricing should be simple, fair, and aligned with value. $0.25 per review. No per-seat fees. That is the entire model.