What Is iRender? How the IaaS Model and Pricing Actually Work
What is iRender and how does its pricing work? iRender is a cloud GPU service that rents you a whole dedicated server rather than taking a job submission. You connect over remote desktop to a machine with 1 to 8x RTX 4090 (24 GB VRAM each), 256 GB of RAM, an AMD Threadripper Pro, NVMe storage, and a 1 Gbps line, install your own software, and render exactly as you would locally. Billing is by the minute while the server runs. The model is summed up by their own line, “Your renders, your rules”: full control, including full responsibility. The control is why scenes match local output, and the responsibility is why the two cost traps, setup time and idle billing, are yours to manage. Both are covered below, along with the bonus programs that change the effective price.
Most render farms ask for your scene. iRender hands you a computer. That single difference explains nearly everything else about the service, including why people who love it love it and why people who get burned get burned. This page lays out how the model works, what the hardware actually is, and what you will really pay, so the rest of our articles can point here instead of repeating it.
Table of Contents
The IaaS Model, in Practice
IaaS means infrastructure as a service. You pick a machine configuration, boot it, and connect over remote desktop to a clean Windows server that is yours for the session. You install your DCC, your render engine, your exact plugin versions, then transfer your project over the 1 Gbps connection and render. Nothing about your pipeline is translated or approximated by someone else’s submission system, which is why output matches your workstation. Files you leave on the server’s storage persist between sessions, so a large asset library or simulation cache uploads once and stays.
The same control cuts the other way. There is no scene checker saving you from a broken file, no support engineer configuring the render for you, and the first session includes a real setup cost, typically 15 to 30 minutes of installing and arranging before the first frame renders. Subsequent sessions are much faster since your environment is saved.
The Hardware
| Component | Spec |
|---|---|
| GPU | RTX 4090, 24 GB VRAM per card, in 1, 2, 4, 6, or 8 GPU configurations |
| CPU | AMD Threadripper Pro |
| RAM | 256 GB |
| Storage | NVMe SSD (2 TB class) |
| Network | 1 Gbps transfer line |
Two notes worth being precise about. The cards are 24 GB each, and multi GPU configurations multiply rendering throughput, not VRAM per frame, so a frame that needs more than 24 GB of contiguous memory still relies on out of core rendering, which the 256 GB RAM pool exists to absorb. And the lineup is RTX 4090, so size your expectations to that card rather than to whatever is newest on a spec sheet.
What It Costs, Before and After the Bonuses
Billing is hourly, metered by the minute, and the rate scales with GPU count. As a reference point, a single 4090 machine lists at roughly $8 per hour [verify current rate before relying on this], with multi GPU configurations priced upward from there. That headline number is not what regular users end up paying, because three programs stack on top of it:
- 100% first deposit bonus. Your first top up is doubled in render credit, which halves the effective rate of a first project.
- Credit Back. A percentage of each session’s spend returns as credit: 10% as standard, 12% during weekday Happy Hours, and 20% across weekends in Golden Hours [verify current tiers and windows].
- Volume bonuses and vouchers. Larger deposits and periodic vouchers add up to 25% more in stages.
Stacked, a regular user’s effective rate lands meaningfully below the rate card, often less than half for someone combining the deposit bonus with weekend rendering. The honest framing is that the list price is the ceiling and the programs decide your real number, so it pays to time heavy sessions to the higher Credit Back windows.
The trap that actually catches people is not the rate, it is the meter. The server bills while it is on, rendering or not. A machine forgotten overnight charges for every idle hour, and on an 8 GPU configuration that is an expensive night’s sleep. Shut the server down the moment the job lands, or set up an auto shutdown script. Every horror story we have heard about iRender pricing traces back to this one behaviour, not to the rates themselves.
Who the Model Fits, and Who It Does Not
iRender fits GPU heavy work where environment control pays off: exact plugin versions, unusual render managers, multi application pipelines, and very large simulation caches that should upload once and stay put. It also fits anyone whose deadline economics favour renting eight 4090s for an afternoon over owning one more card.
It does not fit someone who wants to submit a scene and walk away. That person should use a SaaS farm and pay the convenience premium happily, and our comparison of the four major farms covers which one. There is no shame in either camp. They are different products.
Want to see current configurations and the first deposit bonus?
Frequently Asked Questions
Is iRender a render farm?
Not in the usual sense. A render farm takes your scene and renders it on its own pipeline. iRender rents you a dedicated GPU server you control over remote desktop, with your own software and settings. The result behaves like your workstation with more hardware, which is the appeal for pipeline heavy work and the extra effort for anyone wanting hands off submission.
How much does iRender cost per hour?
A single RTX 4090 machine lists at roughly $8 per hour, with multi GPU configurations priced higher, and billing metered by the minute. The effective rate is usually well below list once the 100% first deposit bonus, Credit Back of 10 to 20%, and volume vouchers stack. Idle time bills at the same rate, so shutting the server down promptly matters more than the rate itself.
Does iRender have GPUs with more than 24GB of VRAM?
No. The lineup is RTX 4090 with 24 GB per card, in configurations up to 8 GPUs. Multiple GPUs increase rendering speed, not the VRAM available to a single frame, so scenes needing more than 24 GB per frame rely on out of core rendering, which the servers’ 256 GB of system RAM is there to support.
See more: Fox vs GarageFarm vs RebusFarm vs iRender: Which Render Farm Fits Your Job?
No comments