Pricing
Midi Coder does not use token-based or AI-usage pricing. This page explains how pricing is built from dedicated infrastructure fees and PAYG usage, and helps teams choose the right tier based on scale, delivery volume, and governance needs.
How Midi Coder Is Priced
Midi Coder uses a Hybrid Dedicated + PAYG model. The most important thing to understand first is that Midi Coder does not charge by token usage and does not charge by raw AI usage. Product pricing is built from two separate but additive components: Dedicated Infrastructure Fee and PAYG Usage Fee.
In short, the monthly cost follows this structure: Monthly total = Dedicated infra fee + PAYG usage fee. In this model, the dedicated infrastructure fee reflects isolated compute capacity for each tenant. PAYG reflects real usage through versions, sub-versions, and sandbox hours.
As for billing timing, the Dedicated infra fee is charged when the project is created. PAYG usage fee is aggregated and billed on a 30-day cycle from the moment the package starts.
Three Infrastructure Tiers And Dedicated Fees
The current pricing model is split across three infrastructure tiers. This dedicated component is used to maintain an isolated EC2 cluster for each tenant. It is not a per-version fee and not an optional add-on. It is the required capacity reservation that activates the PAYG layer above it.
- Tier 1 - Startup: Cloud cost / month
$198.30, Dedicated infra fee$138.80,4 vCPU / 24GB RAM. - Tier 2 - SME: Cloud cost / month
$396.60, Dedicated infra fee$277.60,8 vCPU / 48GB RAM. - Tier 3 - Enterprise: Cloud cost / month
$782.70, Dedicated infra fee$547.80,16 vCPU / 96GB RAM.
According to the current pricing document, the Dedicated infra fee is calculated as 70% × Cloud cost / month. That means when a team chooses a tier, it is first choosing infrastructure capacity, throughput, and the baseline operating envelope of the system before usage-based billing is even applied.
PAYG For Versions, Sub-versions And Sandbox
The PAYG layer is separated into three clear parts: version, sub-version, and sandbox usage.
Version pricing
The complexity of a version is measured through Complexity Score (CS): CS = (#Commands × 2) + (#Queries × 1). From there, the version price follows this formula: Version fee = 8.00 + (CS × 2.00), where Base fee = $8.00 and Unit price = $2.00 per CS.
Version pricing is not split into S / M / L / XL buckets. It increases linearly with CS, which means smaller versions still have a reasonable minimum price while larger versions scale predictably without pricing cliffs. Examples: CS = 10 -> $28.00, CS = 25 -> $58.00, CS = 50 -> $108.00, CS = 80 -> $168.00.
Sub-version pricing
For change requests driven by feedback, the current model uses Sub-version fee = 50% × Version fee. Here, Version fee means the full fee of the original version, including both the base fee and the CS-based component. If a change is only a hotfix and does not create a version, it is not charged.
Sandbox pricing
Sandbox is charged only when it is ON: Tier 1: $1.00 / hour, Tier 2: $1.60 / hour, Tier 3: $2.80 / hour. This applies to flows such as preview, feedback, autofix, and QA. If sandbox is not used, no charge is generated.
Example monthly total
- Dedicated infra:
$277.60 - 6 versions:
$312.00 - 2 sub-versions:
$56.00 - 40 sandbox hours:
$64.00
Total: 277.60 + 312.00 + 56.00 + 64.00 = $709.60. This example shows how the Hybrid Dedicated + PAYG model works in practice: the fixed layer keeps isolated infrastructure available, while the variable layer scales with the amount of real change the team actually delivers.
Which Tier Fits Your Team And Needs
The three current tiers do not differ in core correctness. Every tier keeps the same core principles: Spec = Source of Truth, Code ↔ Contract traceability, and Spec mismatch -> no merge. The differences are instead about infrastructure scale, throughput, level of automation, memory scope, retention, audit / governance, and compliance support.
| Criteria | Tier 1 Startup | Tier 2 SME | Tier 3 Enterprise |
|---|---|---|---|
| Best for | Smaller teams with fewer active users and an earlier-stage delivery rhythm that needs a lighter infrastructure baseline. | Teams delivering steadily and relying more often on preview, sandbox, project collaboration, and feedback loops. | Multi-team or multi-stream organizations that need deeper governance, stronger traceability, and broader compliance coverage. |
| Delivery pace | 3-5 versions / month | 5-10 versions / month | 10-20 versions / month |
| Sandbox | Used only occasionally | Used regularly for preview, QA, and feedback loops | Supports multiple concurrent streams |
| Best when | Feedback on Canvas or Figma pixel-perfect is not yet required | Project-level collaboration and memory are required | Org-level memory, longer retention, and deeper traceability are required |
| Max users | 15 | 30 | 60 |
| Auto-fix loops | 5 | 10 | unlimited |
| Memory scope | none | project-level | org-level |
| Data retention | 30 days | 60 days | 120 days |
| Audit | Code ↔ Contract | Code ↔ Contract ↔ Runtime | Code ↔ Contract ↔ Risk |
The key point is that recommended versions / month should be understood as an operating guideline, not as a hard billing limit.
Pricing FAQ
Does Midi Coder charge by token or raw AI usage?
No. In the current documentation, Midi Coder does not sell tokens and does not sell raw AI usage. Pricing is built around dedicated infrastructure fees plus PAYG for versions, sub-versions, and sandbox hours.
Why is there still a dedicated infra fee if PAYG already exists?
Because each tenant runs on isolated infrastructure. Dedicated infra is the capacity reservation layer, while PAYG reflects actual usage. The two layers are additive, not interchangeable.
Does PAYG version pricing change by tier?
No. The current version formula is the same across all tiers: 8.00 + (CS × 2.00).
What is CS?
CS is Complexity Score, currently calculated as (#Commands × 2) + (#Queries × 1). It reflects contract scope for pricing purposes, not the full end-to-end technical difficulty.
Do larger versions hit pricing cliffs?
No. The current model is linear by CS and does not use S / M / L / XL pricing buckets.
What is the difference between a sub-version and a hotfix?
A sub-version is charged at 50% × Version fee. A hotfix that does not create a version is not charged.
When is sandbox charged?
Only when sandbox is ON. If it is off, no charge is generated.
Which tier fits teams that are just starting?
Smaller or earlier-stage teams generally fit Tier 1 better when version volume is still low and deeper governance layers are not yet required.
Which tier fits governance and enterprise control needs?
Tier 3 is the better fit when the team needs org-level memory, longer retention, deeper auditability, and fuller compliance coverage.
Is there a separate trial, community pricing option, or custom package?
The current pricing document does not explicitly publish a separate trial or community pricing track in this pricing table. For special needs, the safer wording is to direct readers to speak with the team instead of claiming a package that has not been formally published.