Pave vs Upstart (2026)
The build or delegate decision in consumer credit, drawn as sharply as this index can draw it. Pave is a score supplier that publishes a limit on its own authority as a correction: it does not replace a lender's models, it drives lift inside them, and its architecture has no decision output at all, so the lender's underwriting stays the lender's. Its evidence is 100 million evaluations a month and the rare honesty of publishing both sides of the tradeoff, approvals up around 80 percent with defaults down roughly 45 percent, though no lender is named. Upstart is an underwriting partner that takes the decision: more than 90 percent of loans fully automated, retraining daily on loan level repayment outcomes, behind 100+ named and quoted banks and credit unions, 248 state licences and the most extensive regulatory position in this index. It also holds an A on bias disclosure, which almost nothing in this index holds, publishing annual approval and pricing comparisons by race and ethnicity submitted to a federal banking regulator. One keeps your model yours; the other replaces the reason you needed one.
- Your models are an asset you keep. Pave states its products drive lift in a lender's own models and are not a replacement, with no decision output by architecture, so your validated underwriting stack stays yours and lift is measurable inside it.
- Both sides of the tradeoff are published. Approval increases around 80 percent alongside defaults down roughly 45 percent is the pairing that matters, since either number alone can be manufactured by moving a cut off.
- Delivery lands inside your warehouse. Analytics arrive through a cloud data platform's secure sharing under your existing access controls, with the vendor explaining why it abandoned bulk file transfer, a described control rather than an asserted one.
- You want the underwriting outcome, not the tooling. More than 90 percent of loans fully automated across 100+ bank and credit union partners, with the model retrained on 136,000 payments falling due daily, delivers a working consumer lending programme rather than components.
- Fair lending evidence is your gate. A bias A this index has almost never awarded: annual model evaluation published by race and ethnicity and submitted to a federal banking regulator, overseen by named model risk and model governance functions, which almost no credit vendor here evidences.
- Regulatory substance decides it. 248 state lending licences, a published national mortgage identifier, a bank charter application and a Federal Register submission on partner oversight make this a supervised entity, not a software supplier.
This comparison is published by AI FinTech Index, an independent research platform that publishes independent ratings of AI vendors for financial services. Pave and Upstart are each graded against the same capability taxonomy, from each vendor's own public materials and the regulatory record, under the AI FinTech Index verification standard. No vendor pays for placement, and no vendor has reviewed this page. How this evidence is graded
Plain facts
| Pave | Upstart | |
|---|---|---|
| Primary category | Credit Decisioning & Underwriting | Credit Decisioning & Underwriting |
| Founded | Not published | 2012 |
| Headquarters | Not published | San Mateo, California, United States |
| Website | www.pavefi.com | www.upstart.com |
Side by Side
| Axis | P Pave |
U Upstart |
|---|---|---|
| AI Centrality | ||
| Autonomy and Oversight Model | ||
| Model Risk Management and Transparency | ||
| Operational and Outcome Evidence | ||
| AI Safety and Data Stewardship | ||
| GLBA and Data Privacy Posture | ||
| Security Certifications and Trust Center | ||
| Regulatory Status and Licensure | ||
| AI Governance and Bias Disclosure | ||
| AI Liability and Recourse | ||
| Model Supply Chain Disclosure | ||
| Core Systems and Integration Depth | ||
| Deployment Model and Data Residency | ||
| Commercial Transparency | ||
| Institution and Segment Coverage |
The short version of each
Pave
Pave turns a lender's own raw bank transaction data, credit reports and loan performance history into cashflow based scores and attributes, running roughly 100 million evaluations a month. The AI FinTech Index records its defining architectural choice as a limit the company publishes on its own authority: it does not replace a lender's models, it drives lift inside them, and the product has no decision output at all, so the underwriting stays the lender's. It publishes both sides of its performance tradeoff, approvals up around 80 percent with defaults down roughly 45 percent, which is rarer in this category than it sounds, though no lender is named behind the figures. The index flags one unresolved position: assembling and furnishing information bearing on a consumer's creditworthiness is activity United States consumer reporting law defines, and the company sits close to that line without stating where it falls.
Source: AI FinTech Index, 2026
Upstart
Upstart operates an artificial intelligence lending marketplace through which more than 100 named banks and credit unions originate consumer credit, with more than 90 percent of loans completing fully automated, the partner remaining lender of record, and the models retraining daily on loan level repayment outcomes across the whole partner base. The AI FinTech Index records it as holding an A on AI governance and bias disclosure, a grade very few vendors in the index reach, earned by publishing annual approval and pricing comparisons broken out by borrower race and ethnicity and submitting the analysis to a federal banking regulator on the public record. It maintains 248 state lending licences and one of the most extensive regulatory positions the index records. Two things belong beside that: cross partner pooling is the stated mechanism of the product, so every institution's outcomes help price credit for its competitors, and its filings disclose one customer at roughly a third of revenue.
Source: AI FinTech Index, 2026
Common questions
Is Pave better than Upstart for consumer lending?
They sell different things, so better depends on whether you are keeping your credit policy. Pave supplies a score and attribute set that drives lift inside models you continue to own, and its architecture has no decision output at all. Upstart takes the decision, with more than 90 percent of loans fully automated across 100 plus bank and credit union partners while the partner remains lender of record. If you intend to keep your underwriting, Pave. If building and maintaining underwriting is the problem you are shopping for, Upstart. Graded by AI FinTech Index against the same capability axes from each vendor's own published materials, verified August 12, 2026. No vendor pays for placement.
Does either vendor let us keep our own underwriting model?
Pave does by design: the output is a score and attributes applied inside your own credit policy, positioned as additive to conventional assessment. Upstart does not, and is explicit about it. The partner sets credit policy, business objectives and risk appetite, but the model belongs to Upstart and retrains daily on loan level repayment data arriving across the whole partner base. Control at design time is real at Upstart. Control at decision time is not described, with no referral threshold or manual review path published. Graded by AI FinTech Index against the same capability axes from each vendor's own published materials, verified August 12, 2026. No vendor pays for placement.
Does either publish how its model performs across demographic groups?
Upstart does, and it is unusual. It publishes annual approval and pricing comparisons broken out by borrower race and ethnicity against a constructed traditional benchmark, and has submitted that analysis to a federal banking regulator on the public record rather than only as marketing. Three qualifications belong beside it: the benchmark is Upstart's own, approving more applicants at lower rates is a different question from equal error rates across groups, and a model weighing thousands of variables cannot be fully characterised by approval and pricing comparisons. Pave publishes no demographic breakdown. Graded by AI FinTech Index against the same capability axes from each vendor's own published materials, verified August 12, 2026. No vendor pays for placement.
How much do Pave and Upstart cost?
Neither publishes a rate card. Pave publishes both sides of its performance tradeoff, approvals up around 80 percent with defaults down roughly 45 percent, which is rarer than it sounds, but names no lender behind those figures. Upstart is a listed company, so its revenue composition and fee structure appear in periodic filings any prospective partner can read before a sales call, and its filings also disclose one customer at roughly a third of revenue. Neither lets you compare a number without negotiating. Graded by AI FinTech Index against the same capability axes from each vendor's own published materials, verified August 12, 2026. No vendor pays for placement.
Whose data trains the model, and who else benefits from it?
Upstart's pooling is the stated mechanism of the product rather than a side effect: every partner institution's borrower outcomes train the model that prices credit for its competitors. Nothing published states whether a partner may decline to contribute or what becomes of its contribution on exit. Pave's position is different in kind, and its unarticulated boundary is elsewhere: assembling and furnishing information bearing on a consumer's creditworthiness is activity United States consumer reporting law defines, and Pave sits close to that line without ever stating a position on it. Both belong in counsel's review. Graded by AI FinTech Index against the same capability axes from each vendor's own published materials, verified August 12, 2026. No vendor pays for placement.
How does the AI FinTech Index grade Pave and Upstart?
Both are graded on the same fifteen capability axes from public sources, with each grade traceable to the artifact it was read from. The AI FinTech Index records Upstart as holding an A on AI governance and bias disclosure, which very few vendors in the index reach, alongside 248 state lending licences and one of the most extensive regulatory positions recorded here. Pave's distinguishing grade is architectural rather than regulatory: a score supplier that publishes a limit on its own authority. The index publishes no composite score and declares no winner.
Related comparisons
Other published head to head assessments involving these vendors or their closest peers. The full set for this category is on the Credit Decisioning & Underwriting page.
Upstart's cross partner pooling is the stated mechanism of the product, every institution's outcomes train the model pricing credit for its competitors, and its filings disclose one customer at roughly a third of revenue. Pave never articulates its consumer reporting law boundary despite sitting exactly on it, and its Willingness attribute family infers character from spending; both belong in counsel's review.