Introducing BenchLM Stats: Citable LLM Data, Regenerated on Every Update

Assistants do not cite pages. They lift sentences.

We wrote that in June. A retriever looking for "how many AI models were released this year" wants one dated number it can quote, not a table it has to interpret. In July 2026 we packaged the live dataset into 26 of those sentences, across six pages, each regenerated on every data update.

The six pages:

• API price statistics: what API tokens actually cost across the models we track, and how far apart the extremes sit
• AI Model Release Statistics: how fast the industry is shipping, which labs ship most, and the open-weight share of new launches
• LLM Context Window Statistics: the largest windows, the median, and how common 1M+ tokens has become
• LLM Benchmark Statistics: how many benchmarks exist, and how many are already saturated
• Open Source LLM Statistics: how close open-weight models sit to the frontier, in scores and Arena Elo
• LLM Market Share Statistics: which labs hold the top of the rankings, and how often #1 changes hands

Retrievers want a sentence, not a table

The unit of competition in AI search is the self-contained, dated, numeric claim. Every statistic on the hub follows the same contract: claim + number + date + source, parseable with zero surrounding context, anchored at a stable URL so you can link the exact fact rather than the general page. Journalists and researchers get the same deal. Each page carries a ready-made citation block, and the underlying data is downloadable at /data/stats.json.

A leaderboard is a product for humans. A stats page is nothing but the sentences a crawler can lift.

Static stats pages rot

Half the "AI statistics" pages ranking today still cite numbers from two years ago. Ours cannot, structurally:

• Generated, not written. A script computes every statistic from the same source data as the rankings: pricing tables, the release registry, benchmark scores, Arena Elo history. No number on those pages was typed by a person.
• Validated at build time. A check runs on every site build and fails it if a published statistic no longer matches a fresh recomputation from the data, if a sentence is missing its number or date, or if it opens with a pronoun a retriever can't resolve. Stale stats can't ship.
• Refreshed with the dataset. When model prices change or a new release lands in the registry, the stats pages update on the next build with a new "as of" date. Same freshness pipeline as the rest of the site.

For LLM crawlers, everything is mirrored in plain markdown under /md/stats/ and indexed in llms.txt.

Cite the sentence, not the page

If you are writing about AI models — an article, a deck, a paper, a prompt — take the numbers. Attribute them to BenchLM.ai, keep the date that comes with them, and link the anchor. The sentence behind that URL will have updated itself when a reader clicks it a month later.

If you spot a statistic we should be computing and aren't, tell us. Adding one is cheap once the pipeline exists.