FICO 8 vs FICO 9 vs FICO 10: what changed and who uses which

Most US consumers have a single FICO score number in mind — the one their bank app shows them every month — and assume the rest of the FICO universe is a detail. The detail matters, though, because there is no single FICO score. Fair Isaac publishes several generations of the scoring model (FICO 8 from 2009, FICO 9 from 2014, FICO 10 and 10T from 2020) alongside the older mortgage-industry variants (FICO 2, FICO 4, FICO 5) and several industry-specific variants (FICO Bankcard Score, FICO Auto Score). The score your card issuer shows you, the score your auto lender pulls, and the score your mortgage lender pulls are likely three different numbers from three different models on the same underlying data.

This comparison walks through the three modern FICO generations side by side, explains what changed at each model upgrade, and addresses the practical question of which model your lender actually pulls. The answer for most consumer credit cards is still FICO 8; for newer auto and personal loans, often FICO 9; for mortgages, the older FICO 2/4/5 trio (in transition to FICO 10T and VantageScore 4.0 under the Federal Housing Finance Agency’s mandate). The differences across models are usually small — most consumers will see the same approximate score across all three — but for consumers whose files include specific edge cases (medical collections, recently paid collections, rental history, or recent paydowns from high balances), the model matters.

Side-by-side at a glance

Model	Released	Most-used in	Distinctive treatment
FICO 8	2009	Most credit cards, many auto loans, personal loans	Penalizes maxed-out cards heavily, 14-day rate-shopping window, single 30-day-late hits the score sharply.
FICO 9	2014	Some newer auto, personal loan, and credit card programs	Reduces impact of medical collections, ignores paid collections entirely, includes rental data when available, 45-day rate-shopping window.
FICO 10	2020	Selected lenders, growing share	Same baseline as FICO 9; tuned for moderate post-pandemic risk recalibration.
FICO 10T	2020	Newer mortgage products under FHFA transition; some auto	Uses trended data (24-month look-back on balances and payments), penalizing recently paid-down high utilization more than FICO 9.
FICO 2 / 4 / 5	~2003	Current mortgage industry (Equifax = 5, Experian = 2, TransUnion = 4)	Older model; still used because Fannie Mae and Freddie Mac historically required it. Transitioning to FICO 10T + VantageScore 4.0 under FHFA mandate.

FICO 8 — the workhorse model since 2009

FICO 8 was the major scoring overhaul that introduced the published 35/30/15/10/10 weight breakdown most consumers are familiar with, and it has remained the most widely used FICO model across consumer lending for fifteen years and counting. When your credit card issuer shows you a “free FICO score” inside the app, the number is almost always a FICO 8 or a FICO Bankcard Score (which is FICO 8 tuned for credit card decisions specifically). When most auto lenders pull FICO, it is FICO 8 or FICO Auto Score 8. When a personal loan provider underwrites, FICO 8 is the default.

The defining behaviors of FICO 8 are several. It penalizes maxed-out single accounts harshly — a single card at 95% utilization can drag the score 30+ points below the same aggregate utilization spread across multiple cards. It treats medical collections the same as any other collection (which FICO 9 changes substantially). It uses a 14-day rate-shopping window, which is narrower than FICO 9 and FICO 10 (both 45 days) and which can catch out consumers shopping for a mortgage over several weeks.

FICO 8 also includes the “isolated single late payment” treatment, where a consumer with an otherwise long, clean history takes a sharper hit from a single 30-day-late than a consumer with prior delinquencies — because the model interprets the change in behavior as a sharper risk signal. This is the structural reason why a clean 780 file can drop 100+ points from a single missed payment, while a 620 file with several past delinquencies might drop only 50–70 points from the same missed payment.

The dominance of FICO 8 means that for most consumers, the score they see and the score most lenders are using are the same model. The major exceptions are mortgages (still pulling the older FICO 2/4/5 trio) and a growing minority of products pulling FICO 9 or FICO 10.

FICO 9 — medical-debt and rental data, 45-day rate window

FICO 9 was released in 2014 and represented a deliberate effort by Fair Isaac to better predict consumer risk in two specific edge cases — medical collections and consumers who pay rent on time but lack other revolving credit history. The model has been adopted unevenly across the industry; many lenders have stayed on FICO 8 despite FICO 9 being available, both because of regulatory inertia and because the score changes for the average consumer are small.

The two material changes are:

• Medical collections are weighted less. A medical collection in FICO 8 generally hits the score as hard as any other collection. In FICO 9, medical collections receive a substantially reduced penalty, on the rationale that medical debt frequently reflects insurance disputes or billing errors rather than borrower irresponsibility. For a consumer with medical collections on file, the FICO 9 score can be meaningfully higher than the FICO 8 score for the same underlying data — sometimes 20+ points higher.
• Paid collections are ignored. Under FICO 9, a collection account that has been paid off is removed from scoring entirely, regardless of the type. Under FICO 8, the paid collection remains visible and continues to drag the score until the seven-year retention period expires. This change provides a direct incentive for consumers to settle outstanding collections, since the score gain is immediate under FICO 9.

A third change is more contingent. FICO 9 can incorporate rental payment history when the rental data is reported to the bureaus through one of the third-party services (Rental Kharma, RentTrack, etc.) that pushes rental data to TransUnion or Experian. The data is not automatically included; it must be opted into by either the tenant or the landlord. When it is included, on-time rental payments add a positive signal to a consumer who otherwise lacks revolving credit history, which is most useful for younger consumers or recent immigrants building a US credit file.

The rate-shopping window expanded from 14 days (FICO 8) to 45 days (FICO 9) — also carried into FICO 10 — which gives mortgage shoppers and auto loan shoppers a longer window to collect quotes without triggering cumulative inquiry penalty.

FICO 10 and 10T — trended data and post-pandemic recalibration

FICO 10 and FICO 10T were released in early 2020, just before the pandemic disrupted consumer credit behavior. The two models share most behaviors with FICO 9 but introduce one structural change: trended data.

The earlier FICO models (8 and 9) score against a snapshot of your credit file as it exists at the moment the score is calculated. FICO 10T uses a 24-month look-back: the trajectory of your balances and payments over the prior two years, not just the latest data point. The practical effect is that a consumer who has carried high balances historically and only recently paid them down receives a less favorable score in FICO 10T than in FICO 9, because the trended data shows the recent paydown was preceded by a long period of high utilization. Conversely, a consumer with a long history of low utilization who happens to have a high statement balance this month receives a more favorable score in FICO 10T than in FICO 9, because the trended data shows the recent spike is anomalous.

FICO 10 (without the T) is essentially the same baseline as FICO 9 with minor risk-tuning adjustments calibrated to post-2020 consumer behavior. It does not use trended data and behaves close to FICO 9 for most consumers. FICO 10T is the more significant model going forward because the Federal Housing Finance Agency has mandated that the conforming mortgage market transition to FICO 10T (alongside VantageScore 4.0) over a phased timeline starting in 2024–2025.

For the AZEO playbook (covered in the utilization-myth guide), FICO 10T softens the gain. Under FICO 8 or FICO 9, AZEO captures the full per-card and aggregate signal at the moment the score is pulled, because those models only see the snapshot. Under FICO 10T, the trended-data context partly mutes the gain — the model already knows whether you typically run low or high utilization. The AZEO move still helps, but the magnitude is smaller for consumers whose underlying behavior is also low-utilization.

Why your mortgage lender pulls something different

The mortgage industry has historically pulled an older FICO trio — FICO Score 2 (Experian), FICO Score 4 (TransUnion), and FICO Score 5 (Equifax) — collectively called the “Classic FICO” generation, released around 2003 and substantially older than FICO 8. The reason was structural: Fannie Mae and Freddie Mac required these models for conforming mortgages, and most lenders pulled them by default to stay sellable to the GSEs.

The differences between Classic FICO and FICO 8 are not enormous in normal cases, but they can swing 20–50 points for consumers with specific file characteristics — particularly authorized-user accounts (FICO 8 included them by default; older models had different treatment) and recent late payments (the older models scored differently). Consumers preparing for a mortgage application should know which model their lender pulls and run their score through a score simulator calibrated to that model rather than the FICO 8 their card app shows.

The Federal Housing Finance Agency validated FICO 10T and VantageScore 4.0 in 2022, and the transition has since moved forward unevenly. By mid-2025 the agency cleared Fannie Mae and Freddie Mac to begin accepting VantageScore 4.0, while FICO 10T is following on a later timeline — the Enterprises expect to publish historical FICO 10T data in the summer of 2026 and to adopt scores from the model afterward. The FHFA is also shifting the requirement from three credit reports (the long-standing “tri-merge”) to two (“bi-merge”), aligned with the move off Classic FICO. As of 2026 the changeover is a multi-year effort still in progress, and many conforming mortgages are still underwritten on the Classic FICO 2/4/5 trio. Verify with your lender during pre-qualification which model — and how many bureau reports — they will use for the application. (The FHA, separate from the Enterprises, has signaled it will keep tri-merge reporting for now.)

VantageScore — the model Credit Karma shows you

VantageScore is not a FICO model at all. It is a separate scoring product created in 2006 as a joint venture by the three major credit bureaus (Equifax, Experian, TransUnion) to compete with Fair Isaac. The current version, VantageScore 4.0 (released 2017), is the model Credit Karma uses to show consumers their “score” — and is one of the two models (alongside FICO 10T) approved under the FHFA mortgage scoring framework, the first of which Fannie Mae and Freddie Mac were cleared to begin accepting in 2025.

VantageScore 4.0 uses the same 300–850 range as FICO and looks broadly similar, but the underlying algorithm weights factors differently. Utilization is weighted somewhat more heavily in VantageScore than in FICO; new credit somewhat less. The scores you see on Credit Karma can therefore differ from your FICO 8 by 20–50 points in either direction. The differences are mechanical, not errors, and reflect the model differences rather than data discrepancies.

The practical implication is that the score most US consumers see daily (on Credit Karma, on Credit Sesame, on some bank apps) may not be the score the lender uses for the actual decision. Before any major credit pull, verify which model the lender uses and check that specific score if possible — most card issuers offer their own free FICO 8 view, which is closer to what lenders pull than the Credit Karma VantageScore.

Practical implications for the consumer

A few takeaways fall out of the model comparison:

• For most credit card and auto loan decisions, your FICO 8 is the number that matters. The free FICO 8 view from a card issuer (Discover, Citi, Bank of America, Wells Fargo, American Express all offer one) is the closest indicator of what the lender will see.
• For mortgage applications in 2026, the Classic FICO 2/4/5 trio is still the typical pull, but the transition to FICO 10T and VantageScore 4.0 is underway. Confirm with your lender during pre-qualification.
• If you have medical collections on your file, FICO 9 will treat them better than FICO 8 — but most card issuers still pull FICO 8, so the practical benefit depends on whether your prospective lender uses FICO 9.
• If you recently paid down high utilization, FICO 8 and FICO 9 will reward you immediately; FICO 10T will be slower because the trended data shows the prior pattern.
• The score that Credit Karma shows is VantageScore, not FICO. It is useful for tracking trends but is not the number any FICO-using lender will pull.

The differences across models are usually small enough (5–15 points) that they do not change the lender’s decision in most cases. The exception is consumers whose file characteristics are exactly the kind the newer models treat differently — medical collections, paid collections, recent paydowns, rental history. For those consumers, knowing which model the lender uses can be a tier-meaningful detail.

Sources

• FICO 8 / 9 / 10 release dates, treatment of medical collections and paid collections: myFICO — FICO Score versions. Authoritative Fair Isaac source.
• Rate-shopping window per model (14d vs 45d): myFICO — Inquiries on your credit report.
• FICO 10T trended data: Fair Isaac — Introducing FICO Score 10 Suite.
• FHFA mortgage scoring transition: FHFA — Validation and Approval of Credit Score Models.
• VantageScore 4.0 specification: VantageScore — Model versions.

The score-difference figures in this comparison (the 20+ points medical collection swing, the 5–15 typical AZEO gain, etc.) are approximate ranges based on the published model specifications and consumer-simulator output. The actual move for any specific consumer depends on their full file. Use your card issuer’s score simulator or pay for a FICO 9 / FICO 10 view from myFICO to see the specific number a lender on those models would see for your file.