Render your Blender animations without slowing down your workflow
Set up your render in under a minute and keep working while it runs.
Start rendering free10 free credits to start, about 3h 20m of NVIDIA RTX 5090 render time
Rendering shouldn't stop your work
PC blocked for hours
Your machine becomes unusable while rendering, so you can't keep working.
Crashes overnight
Renders fail, freeze, or stop halfway through and you find out the next morning.
Project bottleneck
Rendering eats the time you should be spending on the next shot.
Built for how Blender freelancers actually work
Keep working
Renders run in the background so your PC stays free for the rest of your project.
Reliable output
Pre-flight checks catch issues before they kill an overnight render.
Set up in under a minute
Guided setup with automatic checks. No CLI, no config files.
Only pay for what renders
No subscriptions, no minimums. Billing starts when Blender does and stops when it stops.
Start rendering in under a minute
- 1
Upload
Upload your .blend file straight from the browser.
- 2
Check
Missing textures, HDRIs and plugins are detected automatically.
Clear prompts guide you if anything is missing, before the pod boots.
- 3
Render
Optimized GPU settings are applied and rendering starts instantly on an RTX 5090.
Save presets so future renders are even faster.
- 4
Download
Frames stream back live and can be downloaded any time, while your PC stays free.
See your render in progress
Live preview, real-time stats, and full control from any device.
[frame-upload] frame=3 ms=3812 bytes=2.4 MB
Top-tier GPUs, every render
High-performance hardware without the high cost or setup.
- Powerful GPUs for heavy animation workloads
- Consistent performance across renders
- No hardware limitations holding you back
How fast is the RTX 5090?
FP32 TFLOPS at boost clock. Higher is faster. Every render here runs on a 5090.
What it supports
The full specification, including what we do not run.
Blender versions
Your file renders on the Blender version it was saved with. Every release from 4.4.3 to 5.2.0:
- Blender 4.4.3
- Blender 4.5.0
- Blender 4.5.1
- Blender 4.5.2
- Blender 4.5.3
- Blender 4.5.4
- Blender 4.5.5
- Blender 4.5.6
- Blender 4.5.7
- Blender 4.5.8
- Blender 4.5.9
- Blender 4.5.10
- Blender 4.5.11
- Blender 4.5.12
- Blender 5.0.0
- Blender 5.0.1
- Blender 5.1.0
- Blender 5.1.1
- Blender 5.1.2
- Blender 5.2.0
4.5 is the current LTS. Anything older than Blender 4.4 is refused before it uploads rather than quietly rendered on a version it was not made for, and that exact matching is what keeps geometry nodes behaving the way they do on your machine. New releases are added once an image is built and verified on a real render machine.
Render engines
Both of Blender's built-in engines are supported: Cycles and EEVEE. Cycles renders with OPTIX on the NVIDIA RTX 5090's RT cores and falls back to CUDA if OPTIX is unavailable. Denoising runs on the GPU, including compositor denoise nodes, and persistent data keeps the scene in VRAM between frames.
- Redshift
- V-Ray
- Octane
- LuxCore
The render machine runs stock Blender. A scene set to a third-party engine is flagged before you can launch it, not after you have paid for a machine.
Add-ons and colour management
Nothing is preinstalled, and you are not restricted to a list someone else maintains. Your add-ons are detected in your scene, zipped, and installed on the machine before Blender starts, so an add-on that generates geometry, shading or effects at render time works because you sent it.
Custom OCIO colour configurations are detected and uploaded the same way, which is what stops a render coming back with the wrong exposure. Both are opt-out per job: add-ons that only affect modelling or the interface can be left behind to keep the upload small.
Assets it finds for you
Your .blend is read before anything uploads and every external file it references is listed, so a missing texture is caught before a machine boots rather than after you have paid for one.
- Textures and HDRIs, packed or on disk
- Linked .blend libraries, and their own textures
- Fonts used by text objects
- Audio and video strips, and camera tracking footage
- Alembic and USD caches
- OpenVDB volume sequences
- Mantaflow, cloth, particle and ocean simulation caches
- MDD and PC2 mesh caches
Output formats
PNG sequences come back as PNG, one file per frame, streamed to you as each frame finishes. A scene set to an FFmpeg video renders natively with the codec and container from your file and arrives as a single file at the end. Other image formats currently come back as PNG.
Compositor File Output nodes are repointed onto the render machine automatically, and everything they write is collected afterwards as one zip. That is how multi-layer EXR, render passes and custom AOVs come back today.
Limits and retention
- Frames per job
- 5,000
- Files per job
- 5,000
- Total size per job
- 100 GB
- Single file, browser
- 5 GB
- Single file, add-on
- No practical limit
- Resolution override
- 1 to 500 percent
- GPU memory
- 32 GB VRAM
- Files, frames and logs kept
- 30 days
Uploads go from your machine straight to object storage and are pulled from there by the render machine, so file contents never pass through our web backend. The machine is destroyed when the job ends, along with everything on it.
Every setting explained in full in the user guide.
A whole NVIDIA RTX 5090, to yourself
Every job gets its own NVIDIA RTX 5090 with 32 GB of VRAM. Not a slice of a shared machine, not a position in a queue behind somebody else's job. It boots, pulls your files and starts rendering, and the whole card works on your frames until they are done.
For most people the real comparison is not another farm, it is the card already in their PC. A 5090 has several times the throughput of the cards most artists render on, so a scene that ties up your workstation for a night is a different job here. And your own machine stays completely free the entire time.
- A dedicated GPU per job, never shared
- No queue, rendering starts on boot
- One rate, no priority tiers to choose between
Simple, transparent pricing
One GPU, one rate, no subscription. You can work out what a render costs before you sign up.
1 credit = 20 minutes
A credit buys 20 minutes of render time on an NVIDIA RTX 5090. A frame that takes 4 minutes costs 1 credit.
Billed while Blender runs
Billing starts when Blender starts and stops when it stops. Uploading, downloading and sitting idle cost nothing.
No subscription
No monthly fee, no minimum, no card on file. Buy credits when you need them, and they never expire.
| Pack | Price | Render time | Per render hour |
|---|---|---|---|
| 10 credits | $9.99 | 3h 20m | $3.00 |
| 50 credits | $44.99 | 16h 40m | $2.70 |
| 200 credits | $149.99 | 66h 40m | $2.25 |
Super Renders Farm charges $5.20 per hour for the same RTX 5090, with the same 32 GB of VRAM. We charge $3.00, about 42% less for an identical hour of work.
Their published rate, read from their own pricing page on 1 August 2026. Across five other farms, the same job priced through each farm's own calculator came to $5.02 to $12.84 per render hour on an RTX 4090 (Drop & Render, Render Farm Pricing Compared, 1 April 2026). An hourly rate means nothing without the GPU it was measured on, so we state the card every time. See the full side by side.
Questions people actually ask
Costs, supported versions, what breaks and why.
- How much does it cost to render a Blender animation?
- Which Blender versions does blender.farm support?
- Do my Blender add-ons work on the render farm?
- Can I use blender.farm for paid client work?
- How long will my render take on a render farm?
Or read the full guide, and ask on Discord.
Start your first render for free
10 free credits when you join, which is about 3h 20m of NVIDIA RTX 5090 render time. Enough to run a real project rather than a demo scene.
- No subscription required
- No upfront commitment
- Only pay if you keep using it