Run or Rent

UK · pounds · no account

Is it cheaper to run the model, or rent the API?

For people who use Ollama, LM Studio or llama.cpp, and for small firms deciding whether to buy a GPU or keep paying the API. The sums are in pounds, and every number has a source or is labelled an estimate.

The job
Chatbot for a small teamSummarise 1,000 documentsCoding assistant all day

8 people, 25 questions on each of 21 working days. 1,500 tokens in (the question plus a short history) and 400 tokens out. A worked example, not a log from a real team.

Model and card

from £5,999.99 at Overclockers UK, checked 11 November 2026.

Ofgem price cap, average electricity unit rate for a Direct Debit customer in England, Scotland and Wales on a single-rate default tariff. No VAT on electricity in this period, per Ofgem. The daily standing charge is not included. Your region and payment method can differ. Ofgem, 1 October 2026 to 31 December 2026.

Speed and power

Filled when a sourced run exists. Otherwise type the speed you see in Ollama, LM Studio or llama.cpp.

Starts at the card's maximum rated power. Lower it if you measured the draw.

Rent it

Lean towards renting. The card would take about 16913.6 months to pay for itself on this job. That is a long time to assume the same prices and the same amount of work.

Upper estimate. 575 W × (1900 tokens ÷ 304.9 tokens/sec ÷ 3,600) ÷ 1,000 × 26.32p/kWh

That expression is pence for one job. Divide by 100 for pounds, then multiply by the jobs in the month. Inputs: 575 W maximum rated power, 1,900 tokens, 304.9 tokens/sec, 26.32p/kWh.

API this month
£1.46
Electricity this month
£1.10 · 4.18 kWh
GPU purchase
£5,999.99
Break-even
16913.6 months
  • API cost is input tokens and output tokens, priced per million, times jobs per month. No caching, batch discount, or free tier.
  • Pound figures use the stored exchange rate. They are not a quote from the company that trained the model.
  • Upper estimate. Every token, in and out, is divided by the reported tokens per second. Prompt tokens are usually faster than that, and a card rarely holds its maximum rated power for the whole job, so the real bill is often lower.
  • Formula, in pence per job: 575 W × (1900 tokens ÷ 304.9 tokens/sec ÷ 3,600) ÷ 1,000 × 26.32p/kWh. Divide by 100 for pounds, then multiply by the jobs in the month.
  • The tokens per second are one published run, not a promise for your PC. Context length and software change it.
  • The watts are the card's maximum rated graphics power. Real draw during inference is usually lower.
  • The standing charge on a home bill is not added. A GPU does not create a second one.
  • API price: £0.112 in and £0.447 out per million tokens, from $0.150 and $0.600 at 0.74501 dollars to the pound on 2026-11-10. OpenRouter feed 2026-11-11.
  • Speed: 304.9 reply tokens/sec, 514.5 prompt tokens/sec, Q4_K_M, llama.cpp 0.7.0-dev (commit 1268eda). Run r_vsype4pm1ko, marked Measured by llm-speed. Chat-short. No power draw in the row. Page read 11 November 2026.
  • GeForce RTX 5090: 32 GB GDDR7. Total graphics power: 575 W. Specs read 2026-11-11.
  • Speed is the cited run, not a guess.
  • Upper estimate, based on the official total graphics power of 575 W.

OpenRouter prices · Exchange rate · GPU spec · Speed source · Ofgem unit rate