Tiered Markups Explained: The More You Use, the Lower the Markup

CodeGateway uses tiered pricing based on a 90-day rolling spend: starting at 1.5x and bottoming out at 1.2x. This article explains the five tiers, how the math works, the lock-in mechanic, and answers common questions.

Tiered Markups Explained

TL;DR: Your CodeGateway markup is set by your last-90-day rolling spend in real time, across 5 tiers (1.5× → 1.2×). Use more, pay less per token. No lock-ins, no top-up gimmicks.

CodeGateway uses a pay-as-you-go, save-more-as-you-use pricing model. The markup isn't fixed — it's calculated dynamically based on your accumulated spend over the past 90 days. The more you use, the lower your markup, and the bigger your discount.

This article answers four questions:

  • What are tiered markups, and why design it this way?
  • What are the five tiers?
  • When does the markup change?
  • How can I check my current tier in the Dashboard?

What is a tiered markup

CodeGateway is an official-grade proxy for the Claude API. We give developers in China low-latency, reliable access, but every request costs us upstream tokens paid to Anthropic. The markup is the multiplier we apply on top of Anthropic's official pricing.

For example: Claude Sonnet 4.6's official input price is $3 per 1M tokens. If your markup is 1.5x, you pay $4.5 per 1M tokens for the same request. The markup covers our server costs, monitoring, customer support, and a safety margin to keep things running reliably.

We chose tiers over a single flat markup because the marginal service cost for heavy users is lower — servers, monitoring, and disaster recovery are fixed costs, and amortizing them over more requests pulls the per-unit cost down. Passing those savings back to the developers who actually use the service heavily is the fairer approach.

The five-tier table

Markups are calculated based on your 90-day rolling spend (in USD):

Tier

90-day spend

Markup

Example: tokens per $1 top-up

Tier 1

$0 – $10

1.5x

≈ 667K tokens (Sonnet input)

Tier 2

$10 – $50

1.4x

≈ 714K tokens

Tier 3

$50 – $200

1.3x

≈ 769K tokens

Tier 4

$200 – $500

1.2x

≈ 833K tokens

Tier 5

$500+

1.2x

≈ 833K tokens

Tier 5 is also 1.2x — that's our hard floor, the minimum gross margin needed to keep the service running long-term. It won't go lower.

New users start at Tier 1. As your 90-day rolling spend crosses each threshold, you're automatically promoted to the next tier.

How the 90-day rolling window is calculated

"Rolling" means the window moves forward with time — it isn't fixed from your registration date.

Today: 2026-04-29
Window: 2026-01-29 – 2026-04-29 (the past 90 days)

Tomorrow: 2026-04-30
Window: 2026-01-30 – 2026-04-30

The benefit of this design: long-term steady users always keep their lower markup, and aren't "diluted" by their early trial-period spending. If you stop using the service for a while, the older spend gradually rolls out of the window and your markup may drift back up.

When the markup changes: real-time, based on rolling spend

The markup has nothing to do with top-ups — it's determined by your last 90 days of spend. Every API request looks up your current rolling spend, matches it against the tier table, and bills at that tier's rate.

Example (assuming you just started, so all spend is within the last 90 days):

  • Rolling spend $0: all requests billed at Tier 1 (1.5x)
  • Rolling spend hits $10 (Tier 2 threshold): from the next request onwards billed at Tier 2 (1.4x); already-billed requests are NOT retroactively adjusted
  • Rolling spend hits $50 (Tier 3 threshold): subsequent requests billed at Tier 3 (1.3x), and so on through Tier 5

Common misconceptions:

  • Markup is independent of when you top up — balance is balance, markup is markup, they live in two separate systems
  • There is no lock-in — every request looks up rolling spend in real time
  • Spend more than 90 days old slides out of the window day by day; if you stop using the service, your tier can go back up

Check your current tier in the Dashboard

Log in to the DashboardOverview page. The top of the page shows three numbers:

  • Current tier: determined from your 90-day rolling spend (matches the markup that would be locked on your next top-up)
  • 90-day spend: shown as a dollar amount with a progress bar to the next tier threshold
  • Locked markup for current balance: the actual markup being applied to your current balance

The progress bar is designed so you can see at a glance how much more spend it takes to level up — useful for deciding when to top up next.

FAQ

Q: I'm at Tier 2 with a 1.4x markup. If I stop using it for a month, will I drop back to Tier 1?

A: It depends on how much of your old spend is still inside the 90-day window. If a lot of it has rolled out while you were inactive, yes you'd drop down. But as long as your rolling spend stays at $10 or above, you stay on Tier 2.

Q: When my rolling spend crosses a tier threshold, does the lower markup kick in immediately?

A: Yes. The next API request looks up your rolling spend at request time, finds the new tier, and bills at the new rate. No "lock-in," no delay.

Q: Tier 5 is 1.2x. Will it ever drop further to 1.1x or lower?

A: No. 1.2x is our hard floor (signed off by the CFO) — anything lower and the service can't run sustainably. We'd rather invest the savings into reliability — more regions, lower latency — instead.

Q: Will pricing change in the future?

A: The tier structure and markups are a public commitment, and any change will be announced at least 30 days in advance. When Anthropic adjusts their official pricing, we adjust our base prices in lockstep (the markup itself stays the same).

Related docs

References

AuthorCodeGateway teamReviewed on2026-05-03