Tracy Russell · 2026-7525
PDF text layer (the model received only this text)
SELLER
Tracy Russell Invoice
5442 Riverside Drive
Austin, TX 78739
2026-7525
Tel. (918) 555-0156
Date 09/21/2026
BUYER Terms Net 14
Austin Construction LLC Due date 10/05/2026
2157 Maple Drive
Austin, TX 78781 Pay by ACH
Bank PNC Bank, N.A.
Account name Tracy Russell
Routing number 043000096
Account number 17868931850
Payment reference 2026-7525
Description Qty Unit Unit price Amount
Portland cement, 94 lb bag 35 ea 17.84 624.40
Office chair, ergonomic 4 ea 491.78 1,967.12
No sales tax charged: services are non-taxable. Subtotal 2,591.52
Thank you for your business!
Total 2,591.52
Total due: $2,591.52
Verdict
Cheapest without an error: Gemma 4 31B at $0.17 per 1,000 documents.
- Gemini 3.8 Flash is also error-free but 36.6× 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 | GPT-6 Luna11 / 11 fields | Mistral Small 411 / 11 fields | DeepSeek V4.1 Flash11 / 11 fields | Gemini 3.5 Flash Lite11 / 11 fields | Qwen3.8 Flash10 / 11 fields | Mistral Medium 3.511 / 11 fields | Claude Sonnet 511 / 11 fields | GPT-6 Sol11 / 11 fields | Gemini 3.8 Flash11 / 11 fields |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Supplier | Tracy Russell | Tracy Russell (correct) | Tracy Russell (correct) | Tracy Russell (correct) | Tracy Russell (correct) | Tracy Russell (correct) | Tracy Russell (correct) | Tracy Russell (correct) | Tracy Russell (correct) | Tracy Russell (correct) | Tracy Russell (correct) |
| Supplier ID (EIN, IČO…) | — | — (correct) | — (correct) | — (correct) | — (correct) | — (correct) | — (correct) | — (correct) | — (correct) | — (correct) | — (correct) |
| Invoice number | 2026-7525 | 2026-7525 (correct) | 2026-7525 (correct) | 2026-7525 (correct) | 2026-7525 (correct) | 2026-7525 (correct) | 2026-7525 (correct) | 2026-7525 (correct) | 2026-7525 (correct) | 2026-7525 (correct) | 2026-7525 (correct) |
| Payment reference | 2026-7525 | 2026-7525 (correct) | 2026-7525 (correct) | 2026-7525 (correct) | 2026-7525 (correct) | 2026-7525 (correct) | 2026-7525 (correct) | 2026-7525 (correct) | 2026-7525 (correct) | 2026-7525 (correct) | 2026-7525 (correct) |
| Invoice date | 2026-09-21 | 2026-09-21 (correct) | 2026-09-21 (correct) | 2026-09-21 (correct) | 2026-09-21 (correct) | 2026-09-21 (correct) | 2026-09-21 (correct) | 2026-09-21 (correct) | 2026-09-21 (correct) | 2026-09-21 (correct) | 2026-09-21 (correct) |
| Due date | 2026-10-05 | 2026-10-05 (correct) | 2026-10-05 (correct) | 2026-10-05 (correct) | 2026-10-05 (correct) | 2026-10-05 (correct) | 2026-10-05 (correct) | 2026-10-05 (correct) | 2026-10-05 (correct) | 2026-10-05 (correct) | 2026-10-05 (correct) |
| Net by tax rate | — | 0 %: 2591.52 (correct) | — (correct) | 0 %: 2591.52 (correct) | — (correct) | 0 %: 2591.52 (correct) | : null, %: 2591.52invented value | 0 %: 2591.52 (correct) | 0 %: 2591.52 (correct) | — (correct) | 0 %: 2591.52 (correct) |
| Tax by rate | — | 0 %: 0.00 (correct) | — (correct) | — (correct) | — (correct) | 0 %: 0.00 (correct) | : 0.00, %: 0.00 (correct) | — (correct) | — (correct) | — (correct) | 0 %: 0.00 (correct) |
| Total | 2591.52 | 2591.52 (correct) | 2591.52 (correct) | 2591.52 (correct) | 2591.52 (correct) | 2591.52 (correct) | 2591.52 (correct) | 2591.52 (correct) | 2591.52 (correct) | 2591.52 (correct) | 2591.52 (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 | 043000096 17868931850 | 043000096 17868931850 (correct) | 043000096 17868931850 (correct) | 043000096 17868931850 (correct) | 043000096 17868931850 (correct) | 043000096 17868931850 (correct) | 043000096 17868931850 (correct) | 043000096 17868931850 (correct) | 043000096 17868931850 (correct) | 043000096 17868931850 (correct) | 043000096 17868931850 (correct) |
| Tokens | input / output | 642 / 199 | 645 / 225 | 597 / 149 | 806 / 420 | 630 / 162 | 958 / 3,087 | 597 / 149 | 783 / 174 | 645 / 203 | 630 / 1,527 |
| of which hidden reasoning | billed as output | 0 | 129 | 0 | 311 | 0 | 2,884 | 0 | 0 | 107 | 1,330 |
| of which from cache | input billed at a discount | 312 | 0 | 0 | 442 | 0 | 512 | 0 | 0 | 0 | 0 |
| Cost of this document | this run | $0.0002 | $0.0002 | $0.0002 | $0.0004 | $0.0006 | $0.0015 | $0.0020 | $0.0033 | $0.0033 | $0.0062 |
| Per 1,000 documents | for the same kind of document | $0.17 | $0.18 | $0.18 | $0.36 | $0.59 | $1.53 | $2.01 | $3.31 | $3.32 | $6.20 |
| Latency | this run | 8.8 s | 2.4 s | 1.2 s | 4.8 s | 0.9 s | 47.6 s | 1 s | 2 s | 3.6 s | 5.3 s |