ModelVerdict

Jon Miller · INV250236

English (US)Clean image11 fields

Document en-p0040 · layout en-L04 · 11 fields · public part of the dataset · Same invoice in Czech → · Report an error in this example

Jon Miller · INV250236

Verdict

Cheapest without an error: Gemma 4 31B at $0.16 per 1,000 documents.

  • GPT-6 Sol is also error-free but 95.3× more expensive.

Prompt given to every model

Extract the following data from the attached document and return only JSON with these keys:

supplier_name      supplier's name (exactly as printed, including the legal form)
supplier_id        supplier's tax ID (EIN), if printed
invoice_number     document number
payment_reference  payment reference to quote with the payment
issue_date         invoice date (YYYY-MM-DD)
due_date           due date (YYYY-MM-DD)
net_by_rate        amount per sales tax rate, e.g. {"8.25": 1845.00}; non-taxable amounts under "0"
vat_by_rate        sales tax per rate, e.g. {"8.25": 152.21}
total              total due (number, negative for credit notes)
currency           ISO 4217 currency code (USD, GBP, EUR, …)
bank_account       supplier's bank details as "routing-number account-number", or IBAN

Rules:
- Return the supplier's data only, never the customer's.
- If a value is not on the document, return null. Do not guess.
- If the document has no tax breakdown, return net_by_rate and vat_by_rate as null.
- Amounts as numbers with a decimal point, no thousands separators and no currency symbol.

How we score

Exact match per field after normalising dates, amounts and whitespace. Missing diacritics count as an error. Cost is the real token spend of this run, converted at the rate shown in the footer.

What each model extracted

FieldCorrect valueGemma 4 31B11 / 11 fieldsMistral Small 411 / 11 fieldsDeepSeek V4.1 Flash11 / 11 fieldsGPT-6 Luna11 / 11 fieldsGemini 3.5 Flash Lite11 / 11 fieldsQwen3.8 Flash11 / 11 fieldsGemini 3.8 Flash11 / 11 fieldsMistral Medium 3.510 / 11 fieldsClaude Sonnet 511 / 11 fieldsGPT-6 Sol11 / 11 fields
SupplierJon MillerJon Miller (correct)Jon Miller (correct)Jon Miller (correct)Jon Miller (correct)Jon Miller (correct)Jon Miller (correct)Jon Miller (correct)Jon Miller (correct)Jon Miller (correct)Jon Miller (correct)
Supplier ID (EIN, IČO…)—— (correct)— (correct)— (correct)— (correct)— (correct)— (correct)— (correct)— (correct)— (correct)— (correct)
Invoice numberINV250236INV250236 (correct)INV250236 (correct)INV250236 (correct)INV250236 (correct)INV250236 (correct)INV250236 (correct)INV250236 (correct)INV250236 (correct)INV250236 (correct)INV250236 (correct)
Payment referenceINV250236INV250236 (correct)INV250236 (correct)INV250236 (correct)INV250236 (correct)INV250236 (correct)INV250236 (correct)INV250236 (correct)INV250236 (correct)INV250236 (correct)INV250236 (correct)
Invoice date2025-03-192025-03-19 (correct)2025-03-19 (correct)2025-03-19 (correct)2025-03-19 (correct)2025-03-19 (correct)2025-03-19 (correct)2025-03-19 (correct)2025-03-19 (correct)2025-03-19 (correct)2025-03-19 (correct)
Due date2025-04-032025-04-03 (correct)2025-04-03 (correct)2025-04-03 (correct)2025-04-03 (correct)2025-04-03 (correct)2025-04-03 (correct)2025-04-03 (correct)2025-04-03 (correct)2025-04-03 (correct)2025-04-03 (correct)
Net by tax rate—0 %: 347.85 (correct)0 %: 347.85 (correct)0 %: 347.85 (correct)— (correct)0 %: 347.85 (correct)0 %: 347.85 (correct)0 %: 347.85 (correct)0 %: 347.85 (correct)0 %: 347.85 (correct)— (correct)
Tax by rate—0 %: 0.00 (correct)— (correct)0 %: 0.00 (correct)— (correct)0 %: 0.00 (correct)0 %: 0.00 (correct)0 %: 0.00 (correct)— (correct)— (correct)— (correct)
Total347.85347.85 (correct)347.85 (correct)347.85 (correct)347.85 (correct)347.85 (correct)347.85 (correct)347.85 (correct)347.85 (correct)347.85 (correct)347.85 (correct)
CurrencyUSDUSD (correct)USD (correct)USD (correct)USD (correct)USD (correct)USD (correct)USD (correct)USD (correct)USD (correct)USD (correct)
Bank account091000022 4091383615091000022 4091383615 (correct)091000022 4091383615 (correct)091000022 4091383615 (correct)091000022 4091383615 (correct)091000022 4091383615 (correct)091000022 4091383615 (correct)091000022 4091383615 (correct) 0910000022 4091383615wrong value091000022 4091383615 (correct)091000022 4091383615 (correct)
Tokensinput / output584 / 1922,693 / 1431,549 / 7304,871 / 4291,406 / 1903,108 / 1,6211,406 / 8132,693 / 1445,163 / 1694,871 / 293
of which hidden reasoningbilled as output0061533901,44062300203
of which from cacheinput billed at a discount3070436002,9440000
Cost of this documentthis run$0.0002$0.0005$0.0007$0.0008$0.0009$0.0020$0.0041$0.0051$0.0120$0.0151
Per 1,000 documentsfor the same kind of document$0.16$0.49$0.73$0.82$0.90$2.04$4.10$5.12$12.02$15.11
Latencythis run9 s2.1 s10.1 s6.1 s1.8 s22.8 s5 s2.5 s2.9 s5.9 s