Amy Williams · INV-12943
PDF text layer (the model received only this text)
Amy Williams Invoice
Invoice no. INV-12943
Bill to Invoice date 9/26/25
Dudley-Robles Group Corp. Terms Net 7
7377 Washington Street
Boston, MA 02117 Due date 10/3/25
P.O. number PO-53308
Description Qty Unit Unit price Amount
Safety glasses, clear 97 pr 6.99 678.03
Nitrile gloves, box of 100 2 box 19.00 38.00
Toner cartridge, black 8 ea 75.99 607.92
Drywall sheet 1/2 in. 4 x 8 ft 46 ea 13.99 643.54
Portland cement, 94 lb bag 37 ea 17.63 652.31
Copy paper 8.5 x 11, case of 10 reams 11 case 66.00 726.00
Subtotal 3,345.80
Total 3,345.80
Balance to pay: $3,345.80
No sales tax charged: services are non-taxable.
Thank you for your business!
Amy Williams Pay by ACH
7022 Main Street Bank Bank of
Chicago, IL 60692 America, N.A.
Tel. (934) 555-0194 Account name Amy Williams
Routing number 026009593
Account number 39138564562
Verdict
Cheapest without an error: Gemma 4 31B at $0.18 per 1,000 documents.
- GPT-6 Sol is also error-free but 22.6× 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
| Field | Correct value | Gemma 4 31B11 / 11 fields | Mistral Small 410 / 11 fields | GPT-6 Luna11 / 11 fields | DeepSeek V4.1 Flash11 / 11 fields | Gemini 3.5 Flash Lite11 / 11 fields | Qwen3.8 Flash9 / 11 fields | Mistral Medium 3.511 / 11 fields | Gemini 3.8 Flash11 / 11 fields | Claude Sonnet 511 / 11 fields | GPT-6 Sol11 / 11 fields |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Supplier | Amy Williams | Amy Williams (correct) | Amy Williams (correct) | Amy Williams (correct) | Amy Williams (correct) | Amy Williams (correct) | Amy Williams (correct) | Amy Williams (correct) | Amy Williams (correct) | Amy Williams (correct) | Amy Williams (correct) |
| Supplier ID (EIN, IČO…) | — | — (correct) | — (correct) | — (correct) | — (correct) | — (correct) | — (correct) | — (correct) | — (correct) | — (correct) | — (correct) |
| Invoice number | INV-12943 | INV-12943 (correct) | INV-12943 (correct) | INV-12943 (correct) | INV-12943 (correct) | INV-12943 (correct) | INV-12943 (correct) | INV-12943 (correct) | INV-12943 (correct) | INV-12943 (correct) | INV-12943 (correct) |
| Payment reference | — | — (correct) | PO-53308invented value | — (correct) | — (correct) | — (correct) | PO-53308invented value | — (correct) | — (correct) | — (correct) | — (correct) |
| Invoice date | 2025-09-26 | 2025-09-26 (correct) | 2025-09-26 (correct) | 2025-09-26 (correct) | 2025-09-26 (correct) | 2025-09-26 (correct) | 2025-09-26 (correct) | 2025-09-26 (correct) | 2025-09-26 (correct) | 2025-09-26 (correct) | 2025-09-26 (correct) |
| Due date | 2025-10-03 | 2025-10-03 (correct) | 2025-10-03 (correct) | 2025-10-03 (correct) | 2025-10-03 (correct) | 2025-10-03 (correct) | 2025-10-03 (correct) | 2025-10-03 (correct) | 2025-10-03 (correct) | 2025-10-03 (correct) | 2025-10-03 (correct) |
| Net by tax rate | — | 0 %: 3345.80 (correct) | 0 %: 3345.80 (correct) | — (correct) | — (correct) | 0 %: 3345.80 (correct) | 0 %: 3345.80, : 3345.80, %: 3345.80invented value | — (correct) | 0 %: 3345.80 (correct) | 0 %: 3345.80 (correct) | — (correct) |
| Tax by rate | — | 0 %: 0.00 (correct) | — (correct) | — (correct) | — (correct) | 0 %: 0.00 (correct) | — (correct) | — (correct) | 0 %: 0.00 (correct) | — (correct) | — (correct) |
| Total | 3345.80 | 3345.80 (correct) | 3345.80 (correct) | 3345.80 (correct) | 3345.80 (correct) | 3345.80 (correct) | 3345.80 (correct) | 3345.80 (correct) | 3345.80 (correct) | 3345.80 (correct) | 3345.80 (correct) |
| Currency | USD | USD (correct) | USD (correct) | USD (correct) | USD (correct) | USD (correct) | USD (correct) | USD (correct) | USD (correct) | USD (correct) | USD (correct) |
| Bank account | 026009593 39138564562 | 026009593 39138564562 (correct) | 026009593 39138564562 (correct) | 026009593 39138564562 (correct) | 026009593 39138564562 (correct) | 026009593 39138564562 (correct) | 026009593 39138564562 (correct) | 026009593 39138564562 (correct) | 026009593 39138564562 (correct) | 026009593 39138564562 (correct) | 026009593 39138564562 (correct) |
| Tokens | input / output | 775 / 188 | 729 / 143 | 763 / 420 | 917 / 622 | 763 / 187 | 1,093 / 2,499 | 729 / 127 | 763 / 537 | 949 / 192 | 763 / 262 |
| of which hidden reasoning | billed as output | 0 | 0 | 330 | 519 | 0 | 2,302 | 0 | 350 | 23 | 172 |
| of which from cache | input billed at a discount | 306 | 0 | 0 | 438 | 0 | 512 | 0 | 0 | 0 | 0 |
| Cost of this document | this run | $0.0002 | $0.0002 | $0.0003 | $0.0005 | $0.0007 | $0.0013 | $0.0020 | $0.0026 | $0.0038 | $0.0041 |
| Per 1,000 documents | for the same kind of document | $0.18 | $0.20 | $0.29 | $0.52 | $0.70 | $1.27 | $2.05 | $2.59 | $3.82 | $4.15 |
| Latency | this run | 2 s | 1.1 s | 4.7 s | 9.1 s | 0.9 s | 42.5 s | 1.6 s | 2.2 s | 2.4 s | 4.4 s |