Help·What does generation run on?
What does generation run on?
Dedicated hardware
Real GPUs, not shared cloud instances
Every generation on Endlss runs on dedicated GPU hardware located in Europe. Depending on the model and LoRAs you select, your generation will run on one of the following:
RTX 5090 (32 GB VRAM)
Our primary workhorse for standard image generation. These consumer-flagship GPUs deliver excellent throughput for most models and LoRA combinations at fast turnaround times.
H100 SXM (80 GB VRAM)
Data-centre-grade GPUs used for larger models, video generation, and workloads that require more memory than a single RTX 5090 can provide.
On-demand scaling
Extra capacity when it matters
Our backend orchestration can bring additional hardware online on-demand when load requires it. This includes access to:
H200 SXM (141 GB VRAM)
Next-generation data-centre GPU with significantly more memory, enabling faster processing of the most demanding models.
B200 (180 GB VRAM)
The latest Blackwell-architecture GPU. Reserved for peak demand and future models that push the boundaries of what's possible.
Data residency
Your generations stay in Europe
All generation output — the images and videos you create — is processed and stored on infrastructure within Europe. This means your generated content never leaves European data centres at any point in the pipeline, from the moment the GPU produces it to when it's served back to you.
Why do some models cost more?
Understand how model pricing works and what you get for your credits.
What are AI models?
A guide to every model available on Endlss and what each one does.