onrup

Compute

13 GPU classes.

How does choosing a GPU work?

You choose a class — a family of interchangeable accelerators with the same memory and comparable throughput — rather than a specific machine. Capacity is then sourced as serverless compute across multiple providers. What you decide is memory and speed; where it comes from is not something you have to manage.

Why classes rather than machines

A class is a family of interchangeable accelerators with the same memory and comparable throughput. That is the level at which a decision actually gets made — does the model fit, how fast will it run, what does it cost per hour — and which individual card ends up serving the job changes none of those answers.

Choosing a class rather than a specific machine is also what allows capacity to be found wherever it exists rather than waiting on one queue. The cost of that abstraction is that you cannot pin a region or a host; the benefit is a shorter wait and a lower rate.

Training and serving are priced separately on every class because they are different workloads. Training is bursty and bounded — a run starts, finishes and releases capacity. Serving is resident and continuous, holding capacity available whether or not requests arrive.

ClassVRAMTraining / hrServing / hrGood for
RTX 308012 GB$0.09$0.11Fine-tuning models under 2B parameters · Serving tiny models
RTX 4000 Ada20 GB$0.09$0.11Low-latency serving of small models · Tiny fine-tunes
L424 GB$0.17$0.20Power-efficient inference · Steady-state serving of models up to 8B
RTX 309024 GB$0.22$0.26QLoRA on 7-8B models · LoRA on 3-4B models
RTX 409024 GB$0.38$0.43LoRA on 4-8B models · Multi-adapter serving of an 8B base
A4048 GB$0.42$0.48LoRA on 12-14B models · Preference tuning that needs a reference model · QLoRA on 24B
RTX 6000 Ada48 GB$0.61$0.71Fast 14B LoRA · QLoRA on 24B models · Mixture-of-experts fine-tunes
L40S48 GB$0.78$0.90Multi-adapter serving on an 8B base · Modern 48GB training
A600048 GB$0.65$0.7514B LoRA · Multi-adapter serving · Long-context work at 48GB
A100 40 GB40 GB$1.17$1.3512-14B LoRA · Full fine-tunes of small models · High-bandwidth training
A100 80 GB80 GB$1.68$1.94QLoRA on 70B models · LoRA on 32B models · Long reasoning runs
H100 80 GB80 GB$2.59$2.99Fastest single-GPU runs · fp8 serving · Premium training and serving
H200141 GB$4.55$5.25Mixture-of-experts training · Very long context · Frontier-scale work

Picking one

Two steps. First, filter by memory: the model catalogue publishes an admission threshold per model for half-precision and four-bit training, and a class below that will not be scheduled because the run would fail. Second, choose between the qualifying classes on the trade between hourly rate and throughput.

That second step is where money is won and lost, and it is not always obvious. A faster, dearer class that finishes in half the time costs less overall — which matters most when the cheaper option would require quantised training, because that is meaningfully slower per step.

How much VRAM fine-tuning actually needs →

Last verified 6 August 2026.

Start with the free tier

A magic link creates your account, your tenant and your first API key. No card until you ask for compute.