32 first-party research datasets across the Lattice network, each with a permanent DOI. Direct JSON download. Schema.org Dataset markup on every entry. CC-BY 4.0 - reuse freely with attribution.
Variable lists transcribed for 17 of 32 datasets. Every record carries its full schema at its DOI regardless.
Time-series of input/output token prices for 12 frontier model families across 32 versions, with cost-per-reference-task normalized per entry.
input_price_per_1m_usdoutput_price_per_1m_usdcost_per_reference_task_usdeffective_datecontext_window_tokensPer-tool pricing matrix across the Amazon seller SaaS ecosystem (Helium 10, Jungle Scout, Keepa, et al.) with feature-tier delta and effective monthly cost.
slugnamecategorypriceProfeatureshoursPerDollarratingPer-claim evidence audit of common standing-desk health and productivity marketing claims, sourced to peer-reviewed and meta-analytic literature.
productbrandprice_usdtotalClaimsclaimsSupportedclaimsPartiallySupportedclaimsUnsupportedclaimsContradictedbrandTransparencyScoreclaimsPricing and capability index of 30 go-to-market AI tools across vertical and job-to-be-done, with entry monthly cost, pricing model, and verified-source date.
nameverticaljobentryUsdMonthlypriceModeltierpricingUrldateVerifiedOriginal AI-citation measurement: 20 GTM buying-intent queries across ChatGPT, Perplexity, and Google AI Overviews (60 answers, 162 logged citations), recording which tools each engine names and which sources it cites.
engine tool recommendationscited source URLsdomain citation frequencysource typecross-engine top-tool agreementAI Overview triggerDeployment shape, licence terms and hybrid-search support across vector databases used for retrieval-augmented generation.
OpenTelemetry support, self-host options and pricing models across the LLM observability and operations stack.
AI consumer shrinkflation: what a subscription tier actually delivers over time once limits, quotas and model access are held constant.
Recorded recommendations for AI video generators across engines, counted per engine and per query.
Provider-stated AI API prices, plan tiers, dated price changes and model retirements, each parsed from a saved provider page with its SHA-256, plus a subscribable calendar.
A ledger of incidents that AI API providers declare on their own status pages, one taxonomy across providers, SLA terms as posted, and a revision ledger of records that changed after capture. Declared, not measured.
For each consumer AI plan priced on the provider page, how many tasks of three declared sizes the same money buys at that provider's own API list prices, with the break-even. Provider-stated prices, declared task sizes, arithmetic not advice.
The same fixed corpus counted by the token counters of 13 Claude, 11 Gemini and 2 Cohere models plus offline OpenAI counts, with the ratio between versions. Claude Opus 4.6 to 4.7: English prose +33.3%, Russian, Chinese and Arabic 0%.
A measurement of 11 models on a frozen battery of 60 legitimate prompts, with the protocol frozen before the first call (not lodged with an independent registry). Reported as the protocol said it would: the decay people describe is mostly not visible in refusal, hedging or completeness on this battery. Codes assigned by language-model judges, not validated by humans.
What a finished answer costs, not what a token costs: 9 Claude models, 180 requests in two batteries, 2,160 answers priced at list price. Claude Opus 4.6 to 5.5: list price -20%, cost of a finished answer +38% to +91% depending on the task mix. The output-token change is split into tokenizer, visible writing and hidden thinking.
Where 9 major creator-sales platforms stand on AI-generated content: whether it is allowed, disclosure requirements, monetizability, and detection/enforcement, by modality.
idnamemodalitiesaiAlloweddisclosuredisclosureHowmonetizabledetectionCommercial-use terms, output ownership, copyrightability, and indemnification across 9 AI creative tools, split by free vs paid tier.
idnamemodalityvendorcommercialUseFreecommercialUsePaidownsOutputcopyrightableTimeline of platform and tool policy changes governing AI-generated creative work: date, platform, modality, what changed, impact, and source.
dateplatformmodalitychangeimpactsourceWhere a platform stated creative-rights position contradicts its own terms, scored and cited line by line.
FDA CAERS-derived adverse event reports for dietary supplements over a decade: ingredient category, severity, outcome, year. Modeled estimates.
report_idyearingredient_categoryseverityoutcomeage_bandPharmacy benefit manager spread economics: NADAC reference price vs reimbursed price vs plan-paid price across high-volume generics.
drug_ndcnadac_usdpbm_paid_usdplan_paid_usdspread_usdperiodBrand-name drug patent expirations 2024-2028, generic entry timing, and per-drug savings on substitution.
brand_namegeneric_namepatent_expiry_datefirst_generic_datebrand_price_usdgeneric_price_usdsavings_pctDrugmaker patient assistance rules for 31 drugs at 13 drugmakers, each read on the drugmaker's own page on 2026-10-29: stated one-person income limits run from $31,920 to $79,800; Ozempic shows no figure, Wegovy is not listed, Farxiga was removed for new patients.
State-by-state 1099-K reporting threshold map with federal and state divergence for marketplace sellers and gig-economy operators.
stateabbrregionstate_threshold_2026state_transaction_minpolicy_typegap_vs_federalsource_urllast_verified_datenoteGrants.gov-derived award rate, applicant volume, and median award size across federal grant programs relevant to small business and nonprofits.
program_idagencyapplicantsawardsaward_rate_pctmedian_award_usdPer-niche YouTube CPM and RPM estimates synthesized from creator-shared earnings data, partner studies, and ad-network rate cards.
nichelabelcpm_lowcpm_mediancpm_highrpm_equivalentsample_sizeq4_multipliertop_sourceslast_verified_datePer-course economics across MOOC and bootcamp providers: hours of instruction per dollar, completion rate, and outcome signal.
course_idproviderprice_usdhours_of_instructioncompletion_rate_pctoutcome_signalCost, time-to-complete and reported earnings outcomes across professional certifications, normalised to an hours-per-dollar basis.
Vet-pharmacy vs online-pharmacy pricing for common pet prescription drugs across species and dosage forms.
drug_nameformdose_rangecommon_indicationsvet_price_lowvet_price_highcostco_pricewalmart_pricechewy_pricepetmeds_pricemarkup_multiple_highrequires_vet_scriptidentical_to_human_formulationsource_urlsFont and typeface frequency across recent theatrical-release movie posters and title cards, with genre and decade splits.
titleyeardistributorfont_family_categoryspecific_fontfoundrysurfaceconfidencenoteAll 32 datasets are released under the Creative Commons Attribution 4.0 International license, each with a permanent DOI for stable citation. You may copy, redistribute, remix, transform, and build upon the material for any purpose, including commercially, provided you give appropriate credit, provide a link to the license, and indicate if changes were made.
Suggested citation format for a single dataset:
Couey, V. W. (2026). [Dataset Name]. Deep Synthesis. https://doi.org/10.5281/zenodo.[id]
Licensing, corrections or a custom cut: vincent@deepsynthesis.org
For research-grade citation, prefer the per-study page (each carries full Dataset + ScholarlyArticle JSON-LD with methodology notes).
Preprints, submitted work, in-progress papers across physics and chemistry.
View publicationsCo-bylined branded research at the same methodological standard as the open catalog above. Custom topic, your methodology requirements, distribution included.
Sponsorship inquiry