Printer Search data as of 2026-09-06

Files for NotebookLM

Eight separate documents rather than one long one, because a notebook tool cites separate sources more cleanly than sections of one file. Upload them in the order below and nothing else: JSON and HTML degrade the audio overviews and leak unranked rows into the ranking.

Download the set as one zip, then drag the folder into a new notebook. Or take them one at a time below.

1. 00-source-of-truth.md: the compact authority. Every figure any output may quote, in fixed

tables: knobs, the leaderboard, itemized costs, the head-to-head, sensitivity, dated openness status, known issues, capability flags, data gaps, timing, coverage, the verdict, the glossary. It is generated from the data files and checked against them on every build, and it is the file to tell the notebook to trust when two sources disagree.

1. 01-decision-memo.md: the recommendation, in three pages.
2. 02-shortlist-profiles.md: a standalone profile of each shortlisted machine.
3. 03-methodology-and-coverage.md: who this is for, how it was built, what was searched and

excluded, and the brand-by-brand coverage audit.

1. 04-market-landscape.md: the categories, and what each architecture costs you in practice.
2. 05-comparison-table.csv: one row per scored configuration, for data tables and spreadsheets.
3. 06-sources-and-glossary.md: every source with its date and tier, and every term in plain

language.

1. 07-charts.pdf: cost composition, dimension scores, and rank under each weighting.

Will a notebook tool cope with this?

Comfortably. The set is well inside a single source's limit, so the whole study can sit in one notebook alongside whatever else you put there. Name the notebook with the data stamp (2026-09-06) and rebuild the notebook rather than mixing data versions: every Studio output is a snapshot of the figures it was given.