Local image experiments
€0
Try prompts, models, and looks on the GPU you already own.
Open-weight image models in the RenderBob app. No cloud bill.
Pricing
The app is free. Cloud GPUs are metered.
Run locally at no charge. Pay when you need speed, VRAM, larger models, or a fine-tune.
No pretend-free cloud. You pay for GPU time you actually use.
Typical prices
Experiment on your own hardware at no charge. Pay when you burst a job, run a larger model, or start a fine-tune.
Local app
Local short video loops
€0
Iterate a few seconds of motion without renting a card.
Open video models on your machine. Length and quality follow local VRAM.
Local look exploration
€0
Combine generation with editing until the style holds.
Same studio for models and cuts. Cloud is optional.
Cloud burst
Cloud burst, short clip
Metered GPU
Same workflow as local, on a bigger GPU when the clip is too heavy.
Illustrative: L4 about $0.49/hr, A100 about $1.99/hr. You pay for the run, not the app.
Larger model or batch
Metered GPU
Weights or a queue your tower cannot hold.
Pinned models and look, then extra VRAM and throughput in the cloud.
Fine-tune / own style
Paid cloud
Train a look on cloud GPUs when local hardware is not enough.
Fine-tunes and longer training jobs are billed as cloud GPU time.
Local is always free. Cloud prices depend on GPU class and duration. The numbers below are typical burst jobs, not quotes.
Iterate locally. Burst the job that does not fit.
Most generative work should happen on the machine you already paid for. Cloud is for the pass that needs more VRAM, a bigger model, or a fine-tune.
On your machine
- Daily image and short-video experiments cost nothing beyond electricity.
- You keep the models, the look, and the edit in one app.
- When VRAM or time runs out, the same workflow can move to a cloud GPU.
Same job on RenderBob cloud
You pay for GPU hours, not for a farm subscription. Larger models, faster gens, and fine-tunes are the paid path.
What you pay
The app is free. Cloud GPU time is metered. Team burst work can be scoped as a pin-and-match job when you need ops around a deadline.
If you only burst occasionally, cloud is usually cheaper than buying another card. If you train or run huge batches every day, we will say so.
Pay as you go
- The local app is free. There is no seat fee to generate on your own GPU.
- Cloud GPU time is metered. You pay for the run, not a pretend-free cloud tier.
- Larger models, faster gens, and fine-tunes are paid cloud jobs.
- Teams can pin a look and burst the same workflow without standing up a Comfy server.
GPU hour rates are illustrative for private beta and can change. See the cloud page if you need a scoped burst for a deadline.
Teams and studios
Small teams that share a look can pin models and burst together. We match a local reference, then run on cloud GPUs. This is not a Houdini/Maya pipeline hire.
How RenderBob compares
Your workstation only
Fine for daily experiments. A long video pass, a larger model, or a fine-tune ties up the box you also edit on.
Closed web UIs
Fast to try, locked to their models, and billed whether or not you could have run it locally. RenderBob starts on your machine.
DIY Comfy + GPU rental
You keep the graph, the CUDA nights, and the version drift. RenderBob keeps Comfy-class power under the hood and moves the same look to cloud GPUs when you ask.
Pricing FAQ
Is the local app really free?
Yes. You pay when you use cloud GPUs, larger models, faster gens, or a fine-tune.
Do you give away free cloud GPU time?
No. Beta access to the app can be free. Cloud compute is metered.
What GPU rates should I expect?
Illustrative private-beta rates: L4 about $0.49/hr and A100 about $1.99/hr. The live rate is shown before you confirm a cloud run.
Can I get a VAT invoice?
Yes. Cloud usage produces an EU VAT invoice from TraynMe OÜ.
What if a cloud run fails on your side?
You are not charged for that failed run. Successful outputs stay in the project.
Do I need to learn ComfyUI?
No. Graphs stay under the hood. You can bring a workflow later if you want.