Who builds each model and who actually uses them
FICO scores come from Fair Isaac Corporation, an analytics company that has sold credit risk models to lenders since the 1980s. VantageScore comes from VantageScore Solutions, a joint venture created by the three credit bureaus, Equifax, Experian, and TransUnion, in 2006 to compete with FICO.
Lenders overwhelmingly use FICO. Card issuers, auto lenders, and mortgage lenders pull FICO scores for the vast majority of real lending decisions. Industry estimates have long put FICO's share of lender decisions at roughly 90 percent, and while VantageScore has gained ground, FICO remains the default in underwriting as of early 2026.
VantageScore dominates a different market: free consumer apps. Credit Karma, Chase Credit Journey, Capital One CreditWise, and most other free monitoring tools show VantageScore because it is cheaper to license and the bureaus that own it have an incentive to distribute it. So the score you watch every day and the score that decides your application usually come from different companies.
- ▸FICO: built by Fair Isaac, used in most actual lending decisions
- ▸VantageScore: built by the three bureaus, shown in most free apps
- ▸The score you monitor and the score lenders pull are usually different models
Same range, different math
Both modern model families score from 300 to 850, which makes them look interchangeable. They are not. The inputs overlap heavily, payment history and utilization and age of accounts and inquiries, but the weighting and treatment differ in ways that move real points.
Utilization is the clearest example. FICO 8 looks at your utilization as a snapshot from the most recent statement balances. VantageScore 4.0 and FICO 10T both incorporate trended data, meaning they look at 24 months of balance history to see whether you carry debt or pay in full. A person who runs high statement balances but pays in full every month can look very different under a trended model than under a snapshot model.
Other differences matter at the margins. VantageScore can generate a score with a thinner file, sometimes with as little as one month of history, while classic FICO needs roughly six months. The models also treat paid collections, medical debt, and authorized user accounts somewhat differently, and the exact treatment varies by model version. The honest summary: the two families usually agree on direction, but they can disagree on level by 20 to 50 points or more for any given person.
FICO versions explained
FICO is not one score. It is a family of versions, and lenders adopt new ones slowly because retraining underwriting models is expensive. FICO 8, released in 2009, is still the most widely used version for credit card decisions as of early 2026. When a card issuer pulls your FICO, it is most often FICO 8 from one bureau.
Mortgages run on much older models. Conforming mortgage underwriting has historically required FICO 2 from Experian, FICO 4 from TransUnion, and FICO 5 from Equifax, with the lender using the middle of the three scores. The FHFA has approved a transition to FICO 10T and VantageScore 4.0 for conforming loans, but the rollout has been repeatedly delayed, and the classic scores still drive most mortgage decisions in early 2026. Check current guidance if you are mortgage shopping.
FICO 9 and FICO 10/10T are newer models with friendlier treatment of paid collections and medical debt, plus trended data in 10T. Adoption is real but partial. There are also industry-specific variants, FICO Auto Scores and FICO Bankcard Scores, which run on a 250 to 900 range and weight relevant behavior more heavily. An auto lender quoting your score may be using one of these, which is why it will not match anything you see in an app.
- ▸FICO 8: dominant for credit card decisions
- ▸FICO 2, 4, 5: the classic mortgage trio, one per bureau, middle score wins
- ▸FICO 9 and 10T: newer, kinder to medical debt, 10T adds trended data
- ▸Auto and Bankcard variants: 250 to 900 range, will not match your app
Why your free app score differs from what the lender saw
Three reasons, stacked on top of each other. First, model mismatch: your app probably shows VantageScore 3.0 and the lender probably pulled a FICO version. Second, bureau mismatch: your app shows one or two bureaus, and the lender may have pulled the third. Each bureau has slightly different data on you because not every creditor reports to all three.
Third, timing. Scores recalculate as statements post and balances report. The number your app showed last Tuesday and the number the lender pulled Thursday can differ simply because a balance reported in between.
The practical takeaway: treat your free score as a trend line, not a prediction. If your VantageScore moved up 40 points over six months, your FICO almost certainly improved too. But do not assume the exact number transfers, and do not be surprised when an approval letter quotes a score 30 points away from your app.
Which scores major issuers are known to pull
Issuer pull behavior is not published and varies by applicant state, product, and time, so everything here is a community-observed pattern rather than a guarantee. With that caveat, long-running patterns exist.
Chase is most often reported pulling Experian, with TransUnion and Equifax common in certain states. American Express leans heavily on Experian. Citi frequently pulls Equifax, often alongside Experian. Capital One is famous for pulling all three bureaus on a single application. Bank of America commonly pulls Experian or TransUnion depending on state. Discover and Wells Fargo reports are more mixed.
These patterns matter for two things: knowing which bureau report to clean up before an application, and knowing which freeze to thaw if your reports are frozen. Check recent community data points for your state before relying on any of this, because issuers change bureau relationships without notice.
- ▸Chase: usually Experian, state-dependent
- ▸Amex: usually Experian
- ▸Citi: often Equifax, sometimes plus Experian
- ▸Capital One: typically all three bureaus
- ▸All patterns vary by state and change over time