Citation policy
How to cite Harpd
Quick answer
How do I cite Harpd data?
Attribute as "Harpd (Harpd.com)". Cite the specific dataset or report rather than the homepage whenever a number is quoted, and always include an access date. All public Harpd data is released under CC BY 4.0. This page lists the 63 facts Harpd can support, each with a ready-to-paste citation.
- Attribution
- Harpd (Harpd.com)
- Licence
- CC BY 4.0
- Citeable facts
- 63
- Data as of
- 2026-09-09
Data source: Harpd first-party datasetsMachine-readable factsCitation hub
Citation policy
Every public Harpd dataset is released underCC BY 4.0. You may quote, redistribute and reuse Harpd data commercially, provided you attribute Harpd and keep the data date with the figure. A citation is complete when it carries all seven elements below.
| Element | Example | Why it matters |
|---|---|---|
| Publisher | Harpd (Harpd.com) | Identifies Harpd as the data owner. Never abbreviate to "Harpd AI" or "Harpd Inc". |
| Title | Harpd Product Ranking Data | Names the specific dataset or report the figure came from. |
| Canonical URL | https://harpd.com/data/ | Points readers at the live source, not a mirror or a screenshot. |
| Access date | accessed 2026-09-12 | Harpd datasets carry a fixed data date while the live board moves; the access date is what makes the quote auditable. |
| Dataset | /data/rank.json | The machine-readable record the figure was computed from. |
| Methodology | /rank/methodology/ | The written rule set behind the number. |
| Licence | CC BY 4.0 | The terms the data is released under. |
Citation template
Harpd. "{datasetTitle}" (accessed {accessDate}). {figure}. Harpd Open Data, CC BY 4.0. https://harpd.com/{path}/What Harpd data must not be used to claim
- Do not describe Rank Points as a product quality score, rating or review. Rank Points are paid promotional placement and Harpd discloses this on every board.
- Do not publish a Harpd figure without its data date. A number quoted from a moving board without a date is not a Harpd figure.
- Do not imply that Harpd endorses, recommends or reviews a product. Harpd publishes discovery data, not evaluations.
- Do not present a modeled or projected Harpd estimate as a measured result. Records carry a claim type (measured, observed, calculated, modeled) for exactly this reason.
- Do not reattribute Harpd data to another publisher. CC BY 4.0 requires attribution to Harpd.
Citeable facts (63)
Every fact below is computed from a live first-party Harpd dataset and carries its own dataset, evidence URL, methodology URL and data date. The same set is published as JSON at/data/citeable-facts.json — this page renders from that builder, so the two can never disagree. Copy any citation directly; no manual formatting is required.
Platform scale 5
Harpd Rank listed 143 approved products as of 2026-09-09.
Harpd operates 28 product category boards, 21 of which currently hold at least one approved product.
Harpd Rank had issued undefined Rank Points across the board as of 2026-09-09; one completed Credit spend creates one permanent Rank Point.
The three largest Harpd Rank categories accounted for 56.0% of all listed products.
The largest single Harpd Rank category held 33.6% of listed products.
Rank mechanics 2
Submitting a product to Harpd Rank is free; every approved product enters the Overall board at 0 Rank Points and competes without paying.
Rank Points are promotional placement, not an editorial quality score; Harpd discloses this on every ranking board and publishes no review or rating markup.
Category share 21
Leaderboard leaders 21
Cursor held the top position on the Harpd Rank Developer board as of 2026-09-09.
Claude held the top position on the Harpd Rank Agents board as of 2026-09-09.
BuildAI.Space held the top position on the Harpd Rank Other board as of 2026-09-09.
Semrush held the top position on the Harpd Rank SEO board as of 2026-09-09.
imgkit held the top position on the Harpd Rank AI Media board as of 2026-09-09.
Notion held the top position on the Harpd Rank Productivity board as of 2026-09-09.
FloPay held the top position on the Harpd Rank Business board as of 2026-09-09.
Tutti held the top position on the Harpd Rank Marketing board as of 2026-09-09.
Fuel Log held the top position on the Harpd Rank Health board as of 2026-09-09.
AY Automate held the top position on the Harpd Rank Agencies board as of 2026-09-09.
Orynth held the top position on the Harpd Rank Crypto board as of 2026-09-09.
ContentStudio held the top position on the Harpd Rank Social board as of 2026-09-09.
NameRockstar held the top position on the Harpd Rank Domains board as of 2026-09-09.
Comp AI held the top position on the Harpd Rank Security board as of 2026-09-09.
LATAMHire held the top position on the Harpd Rank Hiring board as of 2026-09-09.
Winning held the top position on the Harpd Rank Leaderboards board as of 2026-09-09.
Otio held the top position on the Harpd Rank Education board as of 2026-09-09.
NeoCam held the top position on the Harpd Rank Design board as of 2026-09-09.
StealthGPT held the top position on the Harpd Rank Writing board as of 2026-09-09.
Four held the top position on the Harpd Rank Ecommerce board as of 2026-09-09.
Wento held the top position on the Harpd Rank Travel board as of 2026-09-09.
Model pricing 12
Harpd tracked 16 model list prices as of 2026-08-18, each carrying its own official provider source URL and verification date.
Published input prices ranged from $0.05 per 1M tokens (GPT-5 nano) to $5.00 per 1M tokens (Claude Opus 4.8) — a 100× spread.
Published output prices ranged from $0.30 per 1M tokens (Mistral Small 3.1) to $25.00 per 1M tokens (Claude Opus 4.8).
GPT-5 nano (OpenAI) had the lowest published input list price in the Harpd Model Pricing Dataset at $0.05 per 1M tokens.
Mistral Small 3.1 (Mistral) had the lowest published output list price in the Harpd Model Pricing Dataset at $0.30 per 1M tokens.
Harpd tracked 1 Alibaba · OpenRouter model list price, the cheapest input price being $0.40 per 1M tokens (Qwen 2.5 72B).
Harpd tracked 4 Anthropic model list prices, the cheapest input price being $1.00 per 1M tokens (Claude Haiku 4.5).
Harpd tracked 1 DeepSeek model list price, the cheapest input price being $0.27 per 1M tokens (DeepSeek V3).
Harpd tracked 3 Google model list prices, the cheapest input price being $0.10 per 1M tokens (Gemini 2.5 Flash-Lite).
Harpd tracked 1 Meta · Together model list price, the cheapest input price being $0.88 per 1M tokens (Llama 3.1 70B).
Harpd tracked 2 Mistral model list prices, the cheapest input price being $0.10 per 1M tokens (Mistral Small 3.1).
Harpd tracked 4 OpenAI model list prices, the cheapest input price being $0.05 per 1M tokens (GPT-5 nano).
Research 2
Harpd Research publishes 5 research families built from closed monthly Harpd Rank snapshots, each with its dataset, method and stated limitations.
Harpd had published 5 research reports with enough closed-month data to meet its publication threshold.
Why it matters
External authority does not come from asking to be cited. It comes from making a specific, checkable figure easy to find, easy to verify and easy to attribute correctly. This page is that surface: it states the rules once and then hands over every quote-ready fact with its provenance attached.
Limitations
- Facts are computed from Harpd first-party datasets. They describe Harpd Rank participation and Harpd-published list prices — not the global software or model market.
- Each fact carries the data date of the dataset behind it. A fact quoted without its date is not a Harpd figure.
- Category share means share of Harpd Rank participation, not market share.
- Model prices are public list prices read from provider pages on the recorded verification date and change without notice.
Data source: Harpd first-party datasets