About codingplans.dev
An independent comparison of AI coding plans, models and enterprise contracts. Every price links the page it came from and the date it was checked.
What the site covers
73 plans, 42 models and 6 enterprise contracts from 33 vendors, the Artificial Analysis Intelligence Index for each model, recommended reasoning effort from FrontierCode 2.0, the DeepSeek peak/off-peak schedule, and 16 briefs on what moves the prices. Changes are logged on the news page.
Where the numbers come from
Each figure comes from the highest source available, in this order:
- The vendor's pricing page or documentation.
- The vendor's announcement.
- Artificial Analysis for benchmarks, speed and cost per task.
- OpenRouter for live token prices.
- Reputable secondary reporting, never for a price the vendor publishes itself.
115 of 115 plan and model records carry a source URL and the date it was checked; the oldest check is 2026-10-24, the newest 2026-10-30. A vendor that does not publish a number gets — and a reason, not an estimate. Benchmarks are copied without rounding.
How the data is checked
The site is generated from public JSON files. Before any page is built, a validator checks every record: required sources and dates, no check dates in the future, annual prices not above monthly prices, and DeepSeek peak rates equal to the published multiple of off-peak. A dataset that fails is not published. Records not rechecked within 180 days are flagged for review.
The same data is free to use: plans.json, models.json, providers.json, apps.json and schedules.json, served with open CORS. A summary for AI assistants is at llms.txt.
The popularity index
The plans table is ordered by an adoption index from 1 to 100 built by this site. It is an estimate of how many people use a plan, not market share and not a quality score.
- Scale. Logarithmic, from 1,000 monthly users (1) to 900 million, ChatGPT's reported weekly active base (100). Doubling reach adds about five points.
- Tier weights. Vendors do not publish how users split across tiers, so a plan's reach is its product's reach times a stated share: free 1.00, entry paid 0.50, mid 0.30, top 0.10, team seat 0.15, enterprise seat 0.04, pay-as-you-go 0.08.
- Confidence. Every plan is labelled: published (14), reported (26), proxy (14), assumed (19). Published means the vendor's own count; reported, a credible third party's; proxy, derived from install or download counts measured on the check date; assumed, an explicit estimate explained in words.
Each plan's signal, source and check date are shown on its vendor page. Treat a two-point gap between proxy rows as noise.
Independence
No vendor pays for placement, ranking or coverage. The site carries no ads and no affiliate or referral links: every outbound link goes straight to the vendor or source. Plans are ordered by the popularity index above, not by vendor. There are no accounts or cookies; page views are counted with privacy-friendly Umami analytics.
Corrections
Spotted an outdated price or a wrong number? Email hello@codingplans.dev with the page and, if you have it, the vendor source. Corrections are made by updating the record in place, with a new check date.