DeepSeek V4.1 Flash vs GPT-6 Sol
Same documents, same prompt: where each model is right, what it really costs and how fast it answers.
At a glance
| Task · documents | English (US) | Czech | German | Spanish | French |
|---|---|---|---|---|---|
| Invoice extraction | 99.4% / 99.9% | 99.9% / 99.9% | 99.5% / 100.0% | 98.3% / 99.2% | 99.4% / 99.9% |
| Email triage | 97.6% / 99.0% | 97.4% / 99.0% | 96.3% / 98.2% | — | — |
| Personal data detection | 99.6% / 99.7% | 99.7% / 99.7% | 99.8% / 99.7% | — | — |
| Contract clauses | 100.0% / 100.0% | 100.0% / 100.0% | 97.2% / 100.0% | — | — |
Correct fields: DeepSeek V4.1 Flash / GPT-6 Sol; the better one in bold. Click a cell to see the details.
Invoice extraction · English (US) (150 documents)
GPT-6 Sol is more accurate by 0.5 pp. DeepSeek V4.1 Flash is 18.6× cheaper.
| Metric | DeepSeek V4.1 Flash | GPT-6 Sol |
|---|---|---|
| Correct fields | 99.4% | 99.9% |
| Error-free documents | 93% | 99% |
| Per 1,000 documents | $0.52 | $9.65 |
| To fix / 1,000 | 67 | 7 |
| Speed (median) | 5.7 sec | 4.1 sec |
| Valid answers | 100.0% | 100.0% |
Accuracy per field
| Field | DeepSeek V4.1 Flash | GPT-6 Sol |
|---|---|---|
| Supplier | 100.0% | 100.0% |
| Supplier ID (EIN, IČO…) | 100.0% | 100.0% |
| Invoice number | 100.0% | 100.0% |
| Payment reference | 96.7% | 100.0% |
| Invoice date | 100.0% | 100.0% |
| Due date | 100.0% | 100.0% |
| Net by tax rate | 100.0% | 100.0% |
| Tax by rate | 100.0% | 100.0% |
| Total | 100.0% | 100.0% |
| Currency | 99.3% | 100.0% |
| Bank account | 97.3% | 99.3% |
Documents where they differ
- Reid Office Products LLC · A-003849 · Scan · DeepSeek V4.1 Flash 10/11, GPT-6 Sol 11/11
- Austin Landscaping Inc. · INV-17836 · Scan · DeepSeek V4.1 Flash 10/11, GPT-6 Sol 11/11
Invoice extraction · Czech (150 documents)
Both are equally accurate. DeepSeek V4.1 Flash is 20.4× cheaper.
| Metric | DeepSeek V4.1 Flash | GPT-6 Sol |
|---|---|---|
| Correct fields | 99.9% | 99.9% |
| Error-free documents | 99% | 99% |
| Per 1,000 documents | $0.50 | $10.12 |
| To fix / 1,000 | 13 | 13 |
| Speed (median) | 5.0 sec | 3.9 sec |
| Valid answers | 100.0% | 100.0% |
Accuracy per field
| Field | DeepSeek V4.1 Flash | GPT-6 Sol |
|---|---|---|
| Supplier | 99.3% | 99.3% |
| Supplier ID (EIN, IČO…) | 100.0% | 100.0% |
| VAT ID | 100.0% | 100.0% |
| Invoice number | 100.0% | 100.0% |
| Payment reference | 100.0% | 99.3% |
| Invoice date | 100.0% | 100.0% |
| Tax point | 100.0% | 100.0% |
| Due date | 100.0% | 100.0% |
| Net by tax rate | 100.0% | 100.0% |
| Tax by rate | 100.0% | 100.0% |
| Total | 100.0% | 100.0% |
| Currency | 100.0% | 100.0% |
| Bank account | 99.3% | 100.0% |
Documents where they differ
- Michaela Strnadová · 25600363 · Scan · DeepSeek V4.1 Flash 12/13, GPT-6 Sol 13/13
Invoice extraction · German (150 documents)
GPT-6 Sol is more accurate by 0.4 pp. DeepSeek V4.1 Flash is 18.8× cheaper.
| Metric | DeepSeek V4.1 Flash | GPT-6 Sol |
|---|---|---|
| Correct fields | 99.5% | 100.0% |
| Error-free documents | 94% | 99% |
| Per 1,000 documents | $0.57 | $10.65 |
| To fix / 1,000 | 60 | 7 |
| Speed (median) | 6.4 sec | 4.9 sec |
| Valid answers | 100.0% | 100.0% |
Accuracy per field
| Field | DeepSeek V4.1 Flash | GPT-6 Sol |
|---|---|---|
| Supplier | 99.3% | 100.0% |
| Supplier ID (EIN, IČO…) | 98.7% | 100.0% |
| VAT ID | 99.3% | 100.0% |
| Invoice number | 100.0% | 100.0% |
| Payment reference | 100.0% | 100.0% |
| Invoice date | 100.0% | 100.0% |
| Tax point | 99.3% | 100.0% |
| Due date | 99.3% | 100.0% |
| Net by tax rate | 100.0% | 100.0% |
| Tax by rate | 100.0% | 100.0% |
| Total | 100.0% | 100.0% |
| Currency | 100.0% | 100.0% |
| Bank account | 98.0% | 99.3% |
Documents where they differ
- Druckerei Baumann GmbH · 2025-0324 · Scan · DeepSeek V4.1 Flash 12/13, GPT-6 Sol 13/13
- Malerbetrieb Otfried Haas · RE-2025-00168 · Scan · DeepSeek V4.1 Flash 12/13, GPT-6 Sol 13/13
- Christoph Haas – Hausmeisterservice · 2025-1432 · Clean image · DeepSeek V4.1 Flash 12/13, GPT-6 Sol 13/13
- Wesermetall GmbH · RE2501375 · Scan · DeepSeek V4.1 Flash 12/13, GPT-6 Sol 13/13
Invoice extraction · Spanish (79 documents)
GPT-6 Sol is more accurate by 0.9 pp. DeepSeek V4.1 Flash is 15.2× cheaper.
| Metric | DeepSeek V4.1 Flash | GPT-6 Sol |
|---|---|---|
| Correct fields | 98.3% | 99.2% |
| Error-free documents | 77% | 89% |
| Per 1,000 documents | $0.71 | $10.76 |
| To fix / 1,000 | 228 | 110 |
| Speed (median) | 5.8 sec | 5.7 sec |
| Valid answers | 100.0% | 100.0% |
Accuracy per field
| Field | DeepSeek V4.1 Flash | GPT-6 Sol |
|---|---|---|
| Supplier | 93.7% | 98.0% |
| Supplier ID (EIN, IČO…) | 100.0% | 100.0% |
| VAT ID | 100.0% | 100.0% |
| Invoice number | 100.0% | 100.0% |
| Payment reference | 97.5% | 100.0% |
| Invoice date | 100.0% | 100.0% |
| Tax point | 98.7% | 100.0% |
| Due date | 100.0% | 100.0% |
| Net by tax rate | 88.6% | 92.0% |
| Tax by rate | 100.0% | 100.0% |
| Total | 100.0% | 100.0% |
| Currency | 100.0% | 100.0% |
| Bank account | 98.7% | 99.0% |
Documents where they differ
- Goñi Logística S.L. · R-2025/1460 · Scan · DeepSeek V4.1 Flash 12/13, GPT-6 Sol 13/13
- Centro de Formación Mediterránea S.L. · 2026/0876 · Scan · DeepSeek V4.1 Flash 12/13, GPT-6 Sol 13/13
- Onofre Diez Guitart · R-2025/1100 · Clean image · DeepSeek V4.1 Flash 12/13, GPT-6 Sol 13/13
- Distribuciones Levante S.L.U. · A-2026-0526 · PDF text · DeepSeek V4.1 Flash 12/13, GPT-6 Sol 13/13
Invoice extraction · French (74 documents)
GPT-6 Sol is more accurate by 0.5 pp. DeepSeek V4.1 Flash is 17.1× cheaper.
| Metric | DeepSeek V4.1 Flash | GPT-6 Sol |
|---|---|---|
| Correct fields | 99.4% | 99.9% |
| Error-free documents | 92% | 99% |
| Per 1,000 documents | $0.65 | $11.05 |
| To fix / 1,000 | 81 | 10 |
| Speed (median) | 5.7 sec | 5.9 sec |
| Valid answers | 100.0% | 100.0% |
Accuracy per field
| Field | DeepSeek V4.1 Flash | GPT-6 Sol |
|---|---|---|
| Supplier | 100.0% | 100.0% |
| Supplier ID (EIN, IČO…) | 100.0% | 100.0% |
| VAT ID | 98.7% | 100.0% |
| Invoice number | 100.0% | 100.0% |
| Payment reference | 98.7% | 99.0% |
| Invoice date | 100.0% | 100.0% |
| Tax point | 100.0% | 100.0% |
| Due date | 100.0% | 100.0% |
| Net by tax rate | 100.0% | 100.0% |
| Tax by rate | 100.0% | 100.0% |
| Total | 100.0% | 100.0% |
| Currency | 100.0% | 100.0% |
| Bank account | 94.6% | 100.0% |
Documents where they differ
- Établissements Bertrand SASU · F2025-0066 · Scan · DeepSeek V4.1 Flash 12/13, GPT-6 Sol 13/13
- Couverture Durand SA · F-11179 · Photo · DeepSeek V4.1 Flash 12/13, GPT-6 Sol 13/13
Email triage · English (US) (150 documents)
GPT-6 Sol is more accurate by 1.4 pp. DeepSeek V4.1 Flash is 10.5× cheaper.
| Metric | DeepSeek V4.1 Flash | GPT-6 Sol |
|---|---|---|
| Correct fields | 97.6% | 99.0% |
| Error-free documents | 79% | 92% |
| Per 1,000 documents | $0.23 | $2.44 |
| To fix / 1,000 | 213 | 80 |
| Speed (median) | 3.8 sec | 2.8 sec |
| Valid answers | 100.0% | 100.0% |
Accuracy per field
| Field | DeepSeek V4.1 Flash | GPT-6 Sol |
|---|---|---|
| Category | 100.0% | 100.0% |
| Priority | 96.7% | 98.0% |
| Sentiment | 82.0% | 93.3% |
| Customer name | 100.0% | 100.0% |
| Order number | 100.0% | 100.0% |
| Invoice number | 100.0% | 100.0% |
| Amount | 100.0% | 100.0% |
| Currency | 100.0% | 100.0% |
| Deadline | 100.0% | 100.0% |
Documents where they differ
- Russell Bates · Email · DeepSeek V4.1 Flash 8/9, GPT-6 Sol 9/9
- Daniel Stewart · Email · DeepSeek V4.1 Flash 8/9, GPT-6 Sol 9/9
- Robert Lewis · Email · DeepSeek V4.1 Flash 8/9, GPT-6 Sol 9/9
- Angela Harmon · Email · DeepSeek V4.1 Flash 8/9, GPT-6 Sol 9/9
Email triage · Czech (150 documents)
GPT-6 Sol is more accurate by 1.6 pp. DeepSeek V4.1 Flash is 10.8× cheaper.
| Metric | DeepSeek V4.1 Flash | GPT-6 Sol |
|---|---|---|
| Correct fields | 97.4% | 99.0% |
| Error-free documents | 77% | 91% |
| Per 1,000 documents | $0.30 | $3.19 |
| To fix / 1,000 | 227 | 93 |
| Speed (median) | 4.2 sec | 2.8 sec |
| Valid answers | 100.0% | 100.0% |
Accuracy per field
| Field | DeepSeek V4.1 Flash | GPT-6 Sol |
|---|---|---|
| Category | 100.0% | 100.0% |
| Priority | 98.7% | 98.0% |
| Sentiment | 78.0% | 92.7% |
| Customer name | 100.0% | 100.0% |
| Order number | 100.0% | 100.0% |
| Invoice number | 100.0% | 100.0% |
| Amount | 100.0% | 100.0% |
| Currency | 100.0% | 100.0% |
| Deadline | 100.0% | 100.0% |
Documents where they differ
- Ignác Bartoš · Email · DeepSeek V4.1 Flash 8/9, GPT-6 Sol 9/9
- Simona Kolářová · Email · DeepSeek V4.1 Flash 8/9, GPT-6 Sol 9/9
- Natálie Kratochvílová · Email · DeepSeek V4.1 Flash 8/9, GPT-6 Sol 9/9
- Václav Vítek · Email · DeepSeek V4.1 Flash 8/9, GPT-6 Sol 9/9
- Šimon Beran · Email · DeepSeek V4.1 Flash 8/9, GPT-6 Sol 9/9
- Růžena Sedláková · Email · DeepSeek V4.1 Flash 8/9, GPT-6 Sol 9/9
Email triage · German (145 documents)
GPT-6 Sol is more accurate by 1.9 pp. DeepSeek V4.1 Flash is 9× cheaper.
| Metric | DeepSeek V4.1 Flash | GPT-6 Sol |
|---|---|---|
| Correct fields | 96.3% | 98.2% |
| Error-free documents | 68% | 84% |
| Per 1,000 documents | $0.30 | $2.73 |
| To fix / 1,000 | 324 | 160 |
| Speed (median) | 4.2 sec | 2.9 sec |
| Valid answers | 100.0% | 100.0% |
Accuracy per field
| Field | DeepSeek V4.1 Flash | GPT-6 Sol |
|---|---|---|
| Category | 100.0% | 100.0% |
| Priority | 97.2% | 97.3% |
| Sentiment | 69.7% | 86.7% |
| Customer name | 100.0% | 100.0% |
| Order number | 100.0% | 100.0% |
| Invoice number | 100.0% | 100.0% |
| Amount | 100.0% | 100.0% |
| Currency | 100.0% | 100.0% |
| Deadline | 100.0% | 100.0% |
Documents where they differ
- Sylwia Gerlach · Email · DeepSeek V4.1 Flash 8/9, GPT-6 Sol 9/9
- Irmela Warmer · Email · DeepSeek V4.1 Flash 8/9, GPT-6 Sol 9/9
- Roselinde Heinz · Email · DeepSeek V4.1 Flash 8/9, GPT-6 Sol 9/9
- Sibille Lehmann · Email · DeepSeek V4.1 Flash 8/9, GPT-6 Sol 9/9
- Nadja Wiek · Email · DeepSeek V4.1 Flash 8/9, GPT-6 Sol 9/9
- Ibrahim Walter · Email · DeepSeek V4.1 Flash 8/9, GPT-6 Sol 9/9
- Margret Löchel · Email · DeepSeek V4.1 Flash 8/9, GPT-6 Sol 9/9
Personal data detection · English (US) (150 documents)
GPT-6 Sol is more accurate by 0.1 pp. DeepSeek V4.1 Flash is 7.3× cheaper.
| Metric | DeepSeek V4.1 Flash | GPT-6 Sol |
|---|---|---|
| Correct fields | 99.6% | 99.7% |
| Error-free documents | 97% | 98% |
| Per 1,000 documents | $0.29 | $2.09 |
| To fix / 1,000 | 27 | 20 |
| Speed (median) | 3.8 sec | 2.6 sec |
| Valid answers | 100.0% | 100.0% |
Accuracy per field
| Field | DeepSeek V4.1 Flash | GPT-6 Sol |
|---|---|---|
| Names | 98.7% | 100.0% |
| Email addresses | 100.0% | 100.0% |
| Phone numbers | 100.0% | 100.0% |
| Addresses | 98.7% | 98.0% |
| Dates of birth | 100.0% | 100.0% |
| National IDs | 100.0% | 100.0% |
| Bank accounts | 100.0% | 100.0% |
Documents where they differ
- Margaret Mills · Text · DeepSeek V4.1 Flash 6/7, GPT-6 Sol 7/7
Personal data detection · Czech (150 documents)
Both are equally accurate. DeepSeek V4.1 Flash is 7.1× cheaper.
| Metric | DeepSeek V4.1 Flash | GPT-6 Sol |
|---|---|---|
| Correct fields | 99.7% | 99.7% |
| Error-free documents | 98% | 98% |
| Per 1,000 documents | $0.31 | $2.22 |
| To fix / 1,000 | 20 | 20 |
| Speed (median) | 4.2 sec | 2.5 sec |
| Valid answers | 100.0% | 100.0% |
Accuracy per field
| Field | DeepSeek V4.1 Flash | GPT-6 Sol |
|---|---|---|
| Names | 100.0% | 100.0% |
| Email addresses | 100.0% | 100.0% |
| Phone numbers | 100.0% | 100.0% |
| Addresses | 98.0% | 98.0% |
| Dates of birth | 100.0% | 100.0% |
| National IDs | 100.0% | 100.0% |
| Bank accounts | 100.0% | 100.0% |
Documents where they differ
Both models got the same number of fields right on every example document.
Personal data detection · German (148 documents)
DeepSeek V4.1 Flash is more accurate by 0.1 pp. DeepSeek V4.1 Flash is 7× cheaper.
| Metric | DeepSeek V4.1 Flash | GPT-6 Sol |
|---|---|---|
| Correct fields | 99.8% | 99.7% |
| Error-free documents | 99% | 98% |
| Per 1,000 documents | $0.31 | $2.13 |
| To fix / 1,000 | 13 | 20 |
| Speed (median) | 3.7 sec | 2.6 sec |
| Valid answers | 100.0% | 100.0% |
Accuracy per field
| Field | DeepSeek V4.1 Flash | GPT-6 Sol |
|---|---|---|
| Names | 100.0% | 100.0% |
| Email addresses | 100.0% | 100.0% |
| Phone numbers | 100.0% | 100.0% |
| Addresses | 98.7% | 98.0% |
| Dates of birth | 100.0% | 100.0% |
| National IDs | 100.0% | 100.0% |
| Bank accounts | 100.0% | 100.0% |
Documents where they differ
- Gunnar Zorbach · Text · DeepSeek V4.1 Flash 6/7, GPT-6 Sol 7/7
Contract clauses · English (US) (150 documents)
Both are equally accurate. DeepSeek V4.1 Flash is 10.4× cheaper.
| Metric | DeepSeek V4.1 Flash | GPT-6 Sol |
|---|---|---|
| Correct fields | 100.0% | 100.0% |
| Error-free documents | 100% | 100% |
| Per 1,000 documents | $0.32 | $3.33 |
| To fix / 1,000 | 0 | 0 |
| Speed (median) | 3.9 sec | 2.5 sec |
| Valid answers | 100.0% | 100.0% |
Accuracy per field
| Field | DeepSeek V4.1 Flash | GPT-6 Sol |
|---|---|---|
| Provider | 100.0% | 100.0% |
| Provider ID | 100.0% | 100.0% |
| Client | 100.0% | 100.0% |
| Client ID | 100.0% | 100.0% |
| Contract type | 100.0% | 100.0% |
| Price (net) | 100.0% | 100.0% |
| Currency | 100.0% | 100.0% |
| Billing period | 100.0% | 100.0% |
| Effective date | 100.0% | 100.0% |
| End date | 100.0% | 100.0% |
| Notice period (months) | 100.0% | 100.0% |
| Late-payment penalty (% per day) | 100.0% | 100.0% |
| Governing law | 100.0% | 100.0% |
Documents where they differ
Both models got the same number of fields right on every example document.
Contract clauses · Czech (150 documents)
Both are equally accurate. DeepSeek V4.1 Flash is 11× cheaper.
| Metric | DeepSeek V4.1 Flash | GPT-6 Sol |
|---|---|---|
| Correct fields | 100.0% | 100.0% |
| Error-free documents | 100% | 100% |
| Per 1,000 documents | $0.42 | $4.62 |
| To fix / 1,000 | 0 | 0 |
| Speed (median) | 4.3 sec | 2.0 sec |
| Valid answers | 100.0% | 100.0% |
Accuracy per field
| Field | DeepSeek V4.1 Flash | GPT-6 Sol |
|---|---|---|
| Provider | 100.0% | 100.0% |
| Provider ID | 100.0% | 100.0% |
| Client | 100.0% | 100.0% |
| Client ID | 100.0% | 100.0% |
| Contract type | 100.0% | 100.0% |
| Price (net) | 100.0% | 100.0% |
| Currency | 100.0% | 100.0% |
| Billing period | 100.0% | 100.0% |
| Effective date | 100.0% | 100.0% |
| End date | 100.0% | 100.0% |
| Notice period (months) | 100.0% | 100.0% |
| Late-payment penalty (% per day) | 100.0% | 100.0% |
| Governing law | 100.0% | 100.0% |
Documents where they differ
Both models got the same number of fields right on every example document.
Contract clauses · German (145 documents)
GPT-6 Sol is more accurate by 2.8 pp. DeepSeek V4.1 Flash is 8.5× cheaper.
| Metric | DeepSeek V4.1 Flash | GPT-6 Sol |
|---|---|---|
| Correct fields | 97.2% | 100.0% |
| Error-free documents | 77% | 100% |
| Per 1,000 documents | $0.51 | $4.36 |
| To fix / 1,000 | 228 | 0 |
| Speed (median) | 4.6 sec | 2.9 sec |
| Valid answers | 100.0% | 100.0% |
Accuracy per field
| Field | DeepSeek V4.1 Flash | GPT-6 Sol |
|---|---|---|
| Provider | 100.0% | 100.0% |
| Provider ID | 81.4% | 100.0% |
| Client | 100.0% | 100.0% |
| Client ID | 82.8% | 100.0% |
| Contract type | 100.0% | 100.0% |
| Price (net) | 100.0% | 100.0% |
| Currency | 100.0% | 100.0% |
| Billing period | 100.0% | 100.0% |
| Effective date | 100.0% | 100.0% |
| End date | 100.0% | 100.0% |
| Notice period (months) | 100.0% | 100.0% |
| Late-payment penalty (% per day) | 100.0% | 100.0% |
| Governing law | 100.0% | 100.0% |
Documents where they differ
- Wesermetall e.K. · Weserform GmbH · Contract text · DeepSeek V4.1 Flash 11/13, GPT-6 Sol 13/13
- Hansawerk GmbH · Schreinerei Weber AG · Contract text · DeepSeek V4.1 Flash 11/13, GPT-6 Sol 13/13
- Schulz Systemtechnik GmbH · Elbdata GmbH · Contract text · DeepSeek V4.1 Flash 11/13, GPT-6 Sol 13/13
- Richter & Scholz e.K. · Köhler AG · Contract text · DeepSeek V4.1 Flash 11/13, GPT-6 Sol 13/13
- Winkler GmbH & Co. KG · Berger & Fuchs e.K. · Contract text · DeepSeek V4.1 Flash 12/13, GPT-6 Sol 13/13
- Walter Systemtechnik GmbH & Co. KG · Weiß Handel GmbH · Contract text · DeepSeek V4.1 Flash 12/13, GPT-6 Sol 13/13
- Möller GmbH & Co. KG · Schreinerei Kaiser GmbH · Contract text · DeepSeek V4.1 Flash 12/13, GPT-6 Sol 13/13
- Fuchs GmbH · Hartmann & Brandt UG (haftungsbeschränkt) · Contract text · DeepSeek V4.1 Flash 12/13, GPT-6 Sol 13/13