Superluminal render farm pricing
By Superluminal Engineering
See the current planning rates and calculate a Blender render estimate from GPU count and elapsed time.
Published 2026-07-29 · Updated 2026-07-29
Superluminal pricing starts at $1 per GPU-hour for one allocated GPU. The rate changes with concurrency, reaching $4 per GPU-hour at 100 GPUs. Your estimate is based on the number of GPUs, elapsed render time, and the rate at that concurrency.
Current planning rates
These are the main points on the current planning curve:
| GPUs used | Rate per GPU-hour | One-hour estimate | Curve point |
|---|---|---|---|
| 1 | $1.00 | $1.00 | Starting rate |
| 5 | $2.00 | $10.00 | 5-GPU rate |
| 25 | $3.00 | $75.00 | 25-GPU rate |
| 100 | $4.00 | $400.00 | 100-GPU rate |
More GPUs can finish independent frames sooner, but speedup depends on the scene, frame distribution, startup work, memory, dependencies, and available capacity.
Calculate an estimate
For a first-pass estimate, use:
estimated render cost = GPUs used × elapsed render hours × rate per GPU-hour at that GPU count
| Scenario | Calculation | Estimated amount |
|---|---|---|
| 1 GPU for 90 minutes | 1 × 1.5 × $1.00 | $1.50 |
| 5 GPUs for 30 minutes | 5 × 0.5 × $2.00 | $5.00 |
| 25 GPUs for 12 minutes | 25 × 0.2 × $3.00 | $15.00 |
| 100 GPUs for 3 minutes | 100 × 0.05 × $4.00 | $20.00 |
These examples exclude tax, credits, promotions, negotiated pricing, partial allocation, and retries. Check the estimate shown in the product before starting the job.
Bundled per-GPU-hour anchor snapshot
Anchor rates rise with concurrency in the reference curve. Total estimated cost also multiplies by GPU count and render time.
| Scenario | Value |
|---|---|
| 1 GPU | $1.00 per GPU-hour |
| 5 GPUs | $2.00 per GPU-hour |
| 25 GPUs | $3.00 per GPU-hour |
| 100 GPUs | $4.00 per GPU-hour |
Why a GPU-hour price is not a project quote
A Blender project has at least two separate variables: how much compute each frame consumes and how effectively its frames can run in parallel. A short animation with heavy frames may need fewer concurrent GPUs for longer. A large animation with independent frames may use more GPUs for a shorter wall-clock interval. Simulation caches, linked files, add-ons, output formats, and memory requirements can restrict the usable node pool.
The best estimate therefore begins with a local sample render. Use a representative frame rather than the easiest frame, record its hardware and Blender version, then separate render time from upload, scene preparation, and download time. Multiply only after deciding how many frames can run independently. If a project has 300 frames and a representative frame takes six minutes on comparable hardware, the raw single-GPU workload is about 30 GPU-hours. That still does not predict a final invoice: farm-side execution and billing rules determine which completed work is chargeable.
Superluminal’s current refund policy provides a support path for suspected farm-side billing faults. That policy is more precise than a blanket promise that every failure mode is automatically free. Keep job identifiers and timestamps when raising a billing question.
What to verify before spending
Before a time-sensitive render, confirm the displayed rate, GPU class, maximum concurrency, available credits or promotions, and the scene’s behavior on a representative frame. If the live product and this page disagree, use the rate shown in the product and ask support to resolve the difference.
The planning rates on this page were checked on 29 July 2026. The complete pricing table is also available as JSON.
Sources
- Superluminal bundled pricing reference artifact
- Superluminal public pricing display
- Superluminal render-farm refund policy