Every software category we track, priced and dated
Software pricing pages are built to be skimmed, not compared. Vendors hide the real seat price behind a billing toggle, rename tiers mid-year, and retire free plans without a changelog entry. ComparEdge pins those numbers down: 494 SaaS, AI and crypto tools across 44 categories, every price read from the vendor's own page and stamped with the date it was checked.
This site is the open-data companion to ComparEdge. The catalog ships as plain JSON feeds anyone can pull, and each category below links to its live ranking, where every tool carries current plans, a free-tier flag and a verification date. The rankings sit in the category leaderboards; the raw plan data lives in the pricing database; and when a tool is not a fit, the alternatives hub maps the switch paths. If a number here ever disagrees with a vendor page, the newer date wins. That is the whole editorial policy.
The 44 categories
AI and machine learning 10 categories · 125 tools
The largest slice of the catalog, and the fastest-moving one. Model pricing shifts more often here than anywhere else, so the LLM entries carry per-token API rates on top of plan data, refreshed on a rolling schedule.
Business and operations 14 categories · 189 tools
The classic stack: CRM seats, email sends, payroll runs. This is where per-seat pricing games are oldest and annual-billing toggles do their best work hiding the real monthly figure.
Crypto and web3 9 categories · 67 tools
Exchanges are priced by maker and taker fee schedules rather than seats, so these categories track full VIP fee tables alongside ordinary subscription products like tax software and portfolio trackers.
Developer and infrastructure 4 categories · 59 tools
Usage-based pricing lives here. Databases and hosting bill by compute unit, storage and egress, which is exactly where a published rate card matters more than a marketing page.
Security and privacy 7 categories · 54 tools
Quote-only pricing is the norm in enterprise security, so verified public numbers are scarce and worth keeping. Where a vendor publishes nothing, the record says so instead of guessing.
Where the catalog is deepest
Coverage is not spread evenly, and that is deliberate. Categories where pricing is most volatile or most opaque get more entries. The ten largest:
Cheapest language models by token price
Language models are priced per million tokens rather than per seat, so the dataset tracks them separately. The table lists the lowest-cost models in the current feed, weighted three parts input to one part output because real workloads read far more than they write. Feed updated 2026-07-10. The same numbers power the LLM cost calculator.
| Provider | Model | Input / 1M | Output / 1M | Context |
|---|---|---|---|---|
| Amazon | Amazon Nova Micro | $0.035 | $0.14 | 128K |
| Cohere | Command R7B | $0.0375 | $0.15 | 128K |
| Cerebras | Llama 3.1 8B | $0.1 | $0.1 | 128K |
| Amazon | Amazon Nova Lite | $0.06 | $0.24 | 300K |
| Groq | GPT-OSS 120B | $0.075 | $0.3 | 128K |
| Groq | Llama 4 Scout | $0.11 | $0.34 | 1000K |
| Gemini 2.5 Flash-Lite | $0.1 | $0.4 | 1000K | |
| DeepSeek | DeepSeek-V4-Flash | $0.14 | $0.28 | 1000K |
| Together AI | Llama 3.1 8B Turbo | $0.18 | $0.18 | 128K |
| Mistral AI | Mistral Small 4 | $0.15 | $0.6 | 128K |
What a plan record looks like
For the rest of the catalog, each record keeps the full plan matrix. The sample below shows the lowest paid monthly plan per product with the free-plan flag and the date the price was last checked against the vendor page. Full profiles are on the tools directory.
| Product | Lowest paid plan | Free plan | Verified |
|---|---|---|---|
| Notion | $12/mo | Yes | 2026-06-28 |
| Linear | $10/mo | Yes | 2026-06-28 |
| Figma | $20/mo | Yes | 2026-06-28 |
| 1Password | $4.99/mo | No | 2026-06-28 |
| HubSpot | $25/mo | Yes | 2026-06-28 |
| Zoom | $16.99/mo | Yes | 2026-06-28 |
| Airtable | $24/mo | Yes | 2026-06-28 |
| Vercel | $20/mo | Yes | 2026-06-28 |
| Mailchimp | $13/mo | Yes | 2026-06-28 |
| GitHub Copilot | $10/mo | Yes | 2026-07-03 |
How the numbers get checked
Every record starts at the vendor's pricing page, rendered in a real browser rather than scraped from cached HTML, because half the interesting numbers only appear after a billing toggle is clicked. A price enters the dataset together with the date it was verified, and that date is part of the data: a figure checked in June is a June figure, not a promise. Changes land in a public changelog feed with the old value, the new value and the day the move was caught. The longer write-up is on this site's methodology page and in the ComparEdge methodology. The short version: no price ships without a human having seen it on the vendor's own page.
Getting the data
Four JSON feeds cover the catalog: llms-tools.json is the product index, llms-pricing.json holds full plan data, llms-changelog.json records price changes over time, and llms-model-prices.json carries per-token model rates. Schema and license are documented on the open data page. If you would rather query from an assistant than parse JSON, three open-source MCP servers wrap the same feeds: the catalog server, llm-cost for token math and price-watch for change tracking. All three are free, need no API key, and are documented on ComparEdge MCP.
The license is CC BY 4.0. Use the data commercially, redistribute it, build on it; just credit ComparEdge. Prices are provided as verified on their stated dates and can move at any time, so treat the verification date as part of the data rather than a footnote.