What 720 code reports tell us about licensing value
A large codebase does not automatically produce a price estimate. We looked at repository size, development history and model outputs to understand what an owner should check before discussing a license.
If you own a private codebase, two questions come first: does it fit what an AI buyer needs, and what could a license be worth? A line count helps describe the asset. The next conversation needs its history, the rights you can grant and a clear account of what is included.
We examined 720 Repo Check report observations dated 19 Oct 2026 to 21 Oct 2026. Of these, 125 contained a stored model estimate. Repeated assessments of the same repository may remain, so these are reports, not 720 verified unique repositories.
The main finding: even among the 185 reports with at least 10,000 original lines, 60 still had no estimate. Use size to frame the review, then examine the evidence behind the range.
An analytical split of the records, not an observed evaluation sequence. Source: Rightmark report snapshot, 22 Oct 2026.
The overall 17.4% rate needs context. Most of this sample consists of small codebases, and none of the 535 reports below 10,000 lines has a stored estimate. That composition pulls down the overall rate.
Within the larger group, 32.4% of reports still have no range. Crossing 10,000 lines therefore does not establish qualification for a license. A missing estimate is a reason to inspect the report; it does not value the code at zero or record a buyer rejection.
Check what the line count actually describes
535 of 720 reports fall below 10,000 original lines. A size comparison is useful only when the report describes the archive you intend to license.
With an estimate / total reports
Source: Rightmark reports-console snapshot, 22 Oct 2026. Original lines include source and test code. The counts beside every bar show reports with an estimate and all reports in the band.
Before placing your project in a band, check that the assessment covers the intended repository and its full history. A partial checkout or a missing component can leave the report describing a different asset from the one you mean to offer.
The goal is an accurate inventory. Original lines here include source and test code. Repo Check runs against a full clone on your own machine; use that preparation step to establish what you actually have.
Larger bands show higher upper estimates
Among the 125 reports with ranges, the median upper estimate rises from $146 to $534 to $1,169. These are model outputs, not prices paid.
Each endpoint is calculated separately within its size band.
Endpoints may come from different reports. They are not a percentile band, a confidence interval or a representative quote. 2 suggested routes are pooled.
| Original lines | Priced reports | Median lower | Median upper |
|---|---|---|---|
| 10,000 to 49,999 | 81 | $14 | $146 |
| 50,000 to 199,999 | 29 | $150 | $534 |
| 200,000+ | 15 | $150 | $1,169 |
This gives an owner a comparison point, not an earnings forecast. The largest band contains only 15 priced reports. 2 suggested licensing routes are pooled, so route mix and other inputs could help explain the difference between bands.
The lower and upper medians are calculated separately and may come from different reports. Historical license duration, exclusivity and estimate confidence are unavailable. Before comparing an estimate with a buyer's offer, establish the data included, the permitted use and the license term.
Development history deserves a closer look
Among the 60 larger reports without an estimate, 37 have fewer than 100 commits and 34 have fewer than 3 contributors; 19 have both characteristics.
These counts describe the unpriced group; they do not fully explain the missing estimates. They do not prove that adding commits or contributors would produce a price. They do suggest a useful next check: whether the history is complete and whether the report accurately reflects how the project was built.
For owners, preserve that development record and gather the evidence of your licensing rights. For buyers, use it to ask more precise questions about the archive behind the proposed price.
Turn the report into a licensing conversation
The practical conclusion is to prepare the asset before negotiating from a number. Start with three things:
We publish Market Notes to make those decisions easier: explain what we are seeing in the data, connect it to the questions buyers ask and show owners what to prepare next.
Private product codebases with full history
Rightmark is assembling a lot from software companies and agencies. We are looking for code your company owns outright, with its full Git history and a test report from your own CI if you have one.
See what qualifies for the lots we are assembling . The code lot is in preparation; it has no published buyer offer.
An indicative standing-rate benchmark
$150 to $1,500
USD per repository, per sale · Medium confidence
This is Rightmark's published benchmark (2026-09), separate from the historical estimates above. Your result can fall outside the range.
Price Index, Q3 2026 · Indicative estimate, not an offer. Final offers depend on the systems you connect, data size and quality, access terms and due diligence.
Start with the repository you want to license
Run Repo Check locally, review share.json, then open that file on the Repo Check page to see the suggested route, fix list and any available estimate. Your source code stays on your machine during this check.
No repository handy? The quick appraisal starts with answers about your code. For questions about fit or the next step, talk to us.