LLM usage

What the transform pipeline spends on the LLM and where it goes. The pipeline runs two to four times a day, so every figure here is a sum over a day's runs rather than a single run — a row read as a day would understate spend by however many runs it dropped.

62 transform runs across 27 days, 2026-08-18 to 2026-09-13

Spend on 2026-09-13
$0.483
2 runs · 3,760 calls
Mean per day
$0.499
Median $0.457 · p90 $0.633 · peak $0.906
Week on week
0%
Last seven days against the seven before them.
2026-09 projected
$14.45
$6.26 over 13 of 30 days, at the current daily rate

Cost per day

A day's spend is the sum of its transform runs. The dashed line is the window mean.

$0$0.25$0.50$0.75$1.00mean $0.49919 Aug24 Aug29 Aug3 Sept8 Sept13 Sept

Cache hit rate

The number that explains the cost above. A run with a cold cache costs several times what the same work costs warm, so a dip here and a spike there are the same event.

0%25%50%75%100%19 Aug24 Aug29 Aug3 Sept8 Sept13 Sept

Cost by call site

Each day's spend split by the stage that made the call. This is what says whether a rise came from more listings needing the LLM or from one stage losing its cache.

$0$0.25$0.50$0.75$1.0020 Aug24 Aug28 Aug1 Sept5 Sept9 Sept13 Sept
  • ask-llm-with-results
  • ask-llm-to-categorise
  • ask-llm
  • identify-multiple-movies
  • identify-shorts

Call sites over the window

Totals across every run collected. Cost per thousand calls is the figure to compare stages by — a stage making many cheap cached calls is not the one to look at first, and a per-call figure would round to zero for all of them.

Call siteCallsCache missesMiss rateCostPer 1k calls
ask-llm-with-results23,00412,74055.4%$8.66$0.376
ask-llm-to-categorise34,17013,26238.8%$1.87$0.055
ask-llm44,55717,89140.2%$1.69$0.038
identify-multiple-movies3,3271,63949.3%$0.814$0.245
identify-shorts7,2973,31445.4%$0.456$0.062

By month

One row per month in the window. Cost per day is the fair comparison between a finished month and one still running.

MonthDaysRunsCallsCachedPrompt tokensCostPer day
2026-08143360,05757.0%63.2M$7.22$0.516
2026-09132952,29856.0%54.4M$6.26$0.482

Biggest prompts

Only the single largest prompt per run is logged, so this is a tally of which venues turn up as the worst offender — not a ranking of every venue's prompt size. A venue here is one whose listings page is large enough to be worth trimming before it reaches the model.

VenueLargest promptRuns toppedCall sites
goethe.de57.9k chars21ask-llm-with-results
bfi.org.uk-southbank52.6k chars40ask-llm-with-results

Venue coverage

How many of the venues in a transform run needed the LLM at all. The rest were handled by the deterministic parsers.

On the most recent run, 190 of 414 venues needed the LLM — 46% of the pipeline.