How does AI used car inspection from photos work?

You upload 3 to 8 photos of a used vehicle — exterior, interior, tires, dashboard, engine bay, undercarriage — and Claude (Anthropic's vision model) examines every photo for visible damage, rust, paint mismatch, tire wear, interior wear, fluid leaks, and modifications. In about 15 seconds you get a structured report: an overall rating, findings grouped by severity, things that look genuinely good, and a list of specific questions to ask the seller.

Can AI really spot damage from photos?

Claude is trained on a very broad distribution of images, including vehicles from every angle and condition. It reliably identifies visible signals an experienced used-car appraiser would notice — paint-texture differences on replaced panels, overspray on trim and glass, rust bubbling through clear-coat, uneven tire wear patterns, curb rash on wheels, sagging headliners, cracked dashboards, fluid leaks under the engine. It cannot see through panels or test mechanical systems, so it is a complement to (not a replacement for) a hands-on pre-purchase inspection by a mechanic.

What photos should I ask the seller for?

Ask for: all four corners of the exterior in daylight, each wheel and tire sidewall, the dashboard with the key on (to capture warning lights), the driver and passenger seats, the trunk floor, the engine bay from above with the hood open, and the undercarriage if possible. Avoid blurry or oblique-angle shots. 3 photos is the minimum; 6-8 gets the best report.

Does this replace a pre-purchase inspection from a mechanic?

No, and you should not skip a mechanic's inspection on any used car you're serious about. What this tool does is help you decide whether the car is worth driving 50 miles to see in the first place. If the AI surfaces severe rust, a cracked windshield across the driver's view, a repainted panel, or mismatched tires before you book the flight or tow truck, that's hours and dollars saved.

Is this free? Do I need to create an account?

Yes — it's free and no account is required. We rate-limit scans to prevent abuse, but the tool is fully open for anyone buying a used car.

Are my photos stored or shared?

Your photos are sent to Claude (Anthropic) for analysis in a single HTTPS request, and the response is returned to you. The photos are not saved to try.vin servers or shared with third parties. Anthropic's API does not use consumer-facing inputs to train models by default.

What can the AI NOT see from photos?

Photos cannot reveal mechanical condition — engine wear, transmission issues, suspension bushings, timing chain stretch, head gasket leaks, electrical gremlins. They also can't show what's under body panels, inside the trunk liner, or behind the interior trim. For any of that you still need a test drive and a mechanic on a lift. The tool will explicitly list "cannot assess" items based on which parts of the car your photos did and did not show.

How accurate is the condition report?

Accuracy is highest on high-contrast, in-focus photos shot in daylight. The tool is conservative by design — it will not report a finding unless a specific visual cue is present, and it calls out when a photo is too dark, blurry, or poorly-angled to evaluate. Every finding includes the photo number it came from so you can verify the AI's reasoning yourself.

How is this different from a VIN check or Carfax?

A VIN check (including try.vin's free decoder) tells you the vehicle's history — title brands, accident records, odometer timeline, open recalls, theft flags. It tells you what has been reported about the car. The AI condition report tells you what the car looks like right now, from the photos you have in front of you. Run both: the VIN history flags a salvage title; the photo inspection flags that the front clip has been replaced with mismatched paint.

Is it safe to buy a car based on photos I haven't seen in person?

Honestly — no. Photos and AI reports reduce risk but they don't eliminate it. The safest path is always: VIN history check, AI photo inspection, video walkaround on a live call with the seller, then a local mechanic's pre-purchase inspection before any money changes hands. Each layer catches what the others miss.

AI · Free · No signup

AI used car inspection from photos alone

Upload 3 to 8 photos of any used car. In about 15 seconds, Claude (Anthropic's flagship vision AI) returns a structured condition report — visible damage, rust, paint mismatch, tire wear, interior wear, and specific questions to ask the seller before you commit to a drive, a deposit, or a flight.

See a sample report

~15sper report

Privatenot stored

Freeno signup

Remoteinspection

Paint mismatch found

Overall: Good

7 photos

Well-kept sedan showing normal wear for its mileage. One moderate paint finding suggests prior body repair…

0 severe 2 moderate 2 minor

Driver fender — paint mismatch

Front tires — outer-edge wear

Front wheel — light curb rash

Driver seat — bolster wear

How the AI inspection works

Three steps, about a minute end to end.

Upload photos

Drop 3–8 photos — exterior, interior, tires, dashboard, engine bay, undercarriage. Any smartphone photo works; we resize automatically.

AI inspects every angle

Claude (Anthropic) examines every photo for visible damage, rust, paint mismatch, tire wear, leaks, modifications, and mileage-vs-wear red flags.

Get a structured report

Overall rating, severity-coded findings with referenced photos, things that look genuinely good, and specific questions to ask the seller.

What the report actually covers

Claude looks for specific visible cues across nine categories — the same things an experienced used-car appraiser checks on a walkaround.

Exterior damage

Dents, scratches, bumper damage, cracked lenses, broken trim.

Rust & corrosion

Surface rust, bubbling paint, rocker rot, rust-through under the body.

Paint mismatch

Different shade or texture on replaced panels, overspray on trim and glass.

Tire wear

Low tread, uneven wear from alignment issues, dry-rot sidewalls, mismatched tires.

Wheels & suspension

Curb rash, bent wheels, visible leaking struts or shocks.

Glass & lights

Cracks in the driver view, hazed headlights, condensation inside lenses.

Interior wear

Seat tears, steering-wheel wear vs claimed mileage, cracked dashboard, staining.

Undercarriage & leaks

Active fluid leaks under engine/transmission, fresh undercoat hiding rust.

Mileage vs wear

Heavy interior wear on a "low-mileage" car — a classic odometer red flag.

Real example

This is what you'll get

A sample report from a used mid-size sedan with ~120k miles. Severity is color-coded; every finding references the photo it came from.

Sample

Overall: Good

Based on 7 photos

Well-kept 2015-era sedan showing normal wear for its mileage. One moderate paint finding suggests a prior body repair on the driver side; front tires are near the end of their service life.

0 severe

2 moderate

2 minor

4 positives

What Claude sees

4 findings, sorted by severity.

Moderate

Paint

Driver-side front fender

Clear-coat orange-peel texture and shade differ slightly from the adjacent door and front bumper. Light overspray visible on the fender-liner plastic inside the wheel arch.

Buyer concern

Likely a prior body repair or repaint on that panel. Worth asking directly about accident history and inspecting the frame rail behind the fender for any straightening marks.

photo 1 photo 2

Moderate

Tires

Front tires (both)

Tread depth visibly low on the outer edges of both front tires. Rear tires look healthier but are a different brand than the fronts.

Buyer concern

Budget $500–900 for a full set of four soon. Uneven front-tire wear on the outer edges often indicates an alignment issue or worn suspension bushings.

photo 4

Minor

Wheels & suspension

Front-passenger alloy wheel

Light curb rash on the outer lip of the front-passenger wheel. No visible bends or cracks.

Buyer concern

Cosmetic only; no impact on drivability. Refinish would cost roughly $100–150 if the buyer cares about appearance.

photo 4

Minor

Interior

Driver's seat outer bolster

Visible upholstery creasing on the outer bolster of the driver seat, consistent with regular entry and exit over 100k+ miles.

Buyer concern

Expected for the mileage. No tears, no cracked leather, no deep staining.

photo 5

Positives

No visible rust on the rocker panels, wheel wells, or any body edges shown.
Headlight lenses are clear — no haze, no condensation inside the housing.
Interior is clean with no stains, tears, or dashboard cracks visible.
All body panel gaps appear even — no signs of major structural repair.

Questions to ask the seller

Bring these up before committing to a test drive or sending a deposit.

Has the driver-side front fender ever been repainted or repaired? Any accident history on that corner?
When were the tires last replaced, and has the front end been aligned recently?
Can I see any service records for suspension work (struts, control arms, bushings)?
Is there documentation of body shop repairs I can review?

Gaps in this report

Photo quality issues

• Photo 6 (engine bay) is slightly underexposed in the corners; some areas of the bay could not be fully assessed.

Not visible in any photo

• Undercarriage — no photo showing the frame, exhaust, or drips below the engine.
• Trunk floor and spare-tire well — not shown, so water-damage signs cannot be checked.

Who this is for

Anyone evaluating a used car they can't physically touch yet.

Buying online

Craigslist, Facebook Marketplace, Cars.com — the car is 200 miles away. Run the seller's photos through try.vin before you book the tow truck.

Out-of-state purchases

Shipping a car sight-unseen? AI pre-inspection is your first line of defense before the auction hammer or the shipping-company pickup.

First-time buyers

You don't know what a resprayed fender looks like — but Claude does. Get a second opinion from someone who has seen millions of car photos.

Vetting dealer listings

Before you drive across town for a test drive, let AI tell you what the listing photos actually reveal — including the things the dealer didn't point out.

Why this actually works

Modern vision AI sees the things an experienced buyer would notice — and it doesn't get tired, distracted, or emotionally invested in the sale.

~15 seconds

Full report in the time it takes to boil a kettle.

Photos not stored

Analyzed on the fly, never saved to try.vin or used for training.

Free, no signup

Just the tool. No account, no email, no credit card.

Claude Sonnet 4.6

Anthropic's flagship vision model, not generic OCR.

Your turn

Inspect your car now

Drop 3 to 8 photos below. Your first report will be ready before you finish reading the FAQ.

Upload photos

3–8 photos · JPEG, PNG, or WebP · up to 5 MB each. More angles = more accurate report.

Drop photos here, or click to choose

0 / 8 photos added

Tell us about the car (optional, improves accuracy)

Mileage is especially useful — it lets Claude flag "worn interior on a low-mileage car" as a potential odometer-rollback red flag.

Year

Make

Model

Mileage

Anything else Claude should know?

Photos are analyzed by Claude (Anthropic) and are not stored. This tool is a visual assessment — always get a pre-purchase inspection from a qualified mechanic before buying.

Frequently asked questions

Short answers to the questions used-car buyers ask about AI photo inspection.

Don't drive 200 miles for a bad car.

Spend 60 seconds first. Drop the seller's photos into try.vin and let Claude tell you what's really there before you commit to the drive, the flight, or the deposit.