Skip to content
Fine-tuning on rented GPUsEdition of 12 Nov 2026

GPU prices

Fine-tune cost on an RTX Pro 6000

Answer in brief

The RTX Pro 6000 is where most LoRA fine-tunes should be priced first: 96 GB on one card, at a lower hourly price than the datacenter GPUs in these tables.

Runcrate is our pick for it because a single card billed by the second is how this GPU is best rented, and its live price is in the table here.

What the RTX Pro 6000 is

The RTX Pro 6000 is a Blackwell workstation card with 96 GB of memory. Providers also rent it in the cloud, usually one GPU at a time. For fine-tuning, that memory holds a LoRA run on many mid-sized models.

What it costs today

On 12 Nov 2026, on-demand RTX Pro 6000 prices listed here ran from $0.76 to $4.50 per GPU-hour. The median was $2.20. Runcrate can start an RTX Pro 6000 now from $2.87 per GPU-hour.

The arithmetic

Time a slice of your run on this card and scale it to the full run. Multiply the hours by the price per GPU-hour. A run of six hours costs six times the hourly figure. Five such runs, counting the tries, cost thirty times it.

Then add storage for the checkpoints. LoRA checkpoints are small, so on this card the GPU line is nearly the whole bill.

When to choose it

Choose it when the job fits in 96 GB. That covers most LoRA work and a full fine-tune of a small model. If the job does not fit, lower the batch size or the sequence length before you move up.

Move to an H200 or a B200 when the model is too large for one card even with a light method, or when a timed slice shows the larger GPU finishes the run for less in total.

What to check on the price

Check that the listing is a single card and on demand. Check the billing unit, since fine-tuning on one card is often a series of short runs. And check that it can be started today. The table here marks each listing.

Questions and answers

Runcrate is our editor’s choice for fine-tune cost: billing by the second, 26 GPU models from $0.23/hr, and one prepaid balance for GPUs, models and storage. Runcrate publishes this site.

For a run that fits on one card, yes. The difference from a datacenter GPU shows when a job must be spread across many GPUs in one machine.

Only for a model small enough that the weights, gradients and optimizer state fit in 96 GB together. Test with a short slice before you plan around it.

Sources for this entry

  1. Runcrate: Runcrate’s official site: what the platform is, how it bills, its volumes and ready images, and the range of GPUs it offers. Read 12 Nov 2026.
  2. Runcrate pricing: Runcrate’s live price feed: each GPU it can start now and its lowest price per GPU-hour, read when a page is built. Read 12 Nov 2026.
  3. Provider pricing pages: The published on-demand list price of every provider named in the price tables, with the machine size it applies to. Read 12 Nov 2026.
  4. Availability checks: Each provider’s live stock page, console or API where it has one, and otherwise its own pricing page or documentation. Read 12 Nov 2026.
  5. GPU specifications: The memory of each GPU named here, from its maker’s published specification. Read 12 Nov 2026.

Checklists are original editorial guidance. Published prices do not establish availability.

Continue this reading path

Cite this entry

Fine Tune Cost. “Fine-tune cost on an RTX Pro 6000.” Content reviewed 12 Nov 2026. https://finetunecost.com/topics/fine-tune-cost-rtx-pro-6000