ModelVerdict

Comparison

GPT-6 Luna vs Qwen3.8 Flash

Same documents, same prompt: where each model is right, what it really costs and how fast it answers.

At a glance

Task · documentsEnglish (US)CzechGermanSpanishFrench
Invoice extraction99.7% / 82.4%99.3% / 91.8%99.2% / 93.5%99.1% / 87.9%98.9% / 90.6%
Email triage97.5% / 97.5%97.9% / 98.5%97.0% / 97.9%——
Personal data detection99.6% / 99.2%99.6% / 97.0%99.6% / 99.6%——
Contract clauses99.8% / 95.1%100.0% / 99.9%100.0% / 98.0%——

Correct fields: GPT-6 Luna / Qwen3.8 Flash; the better one in bold. Click a cell to see the details.

Invoice extraction · English (US) (150 documents)

GPT-6 Luna is more accurate by 17.3 pp. GPT-6 Luna is 2.3× cheaper.

MetricGPT-6 LunaQwen3.8 Flash
Correct fields99.7%82.4%
Error-free documents97%57%
Per 1,000 documents$0.51$1.15
To fix / 1,00033427
Speed (median)3.4 sec17.4 sec
Valid answers100.0%90.7%

Accuracy per field

FieldGPT-6 LunaQwen3.8 Flash
Supplier100.0%85.3%
Supplier ID (EIN, IČO…)100.0%88.7%
Invoice number99.3%85.3%
Payment reference98.0%79.3%
Invoice date100.0%85.3%
Due date100.0%85.3%
Net by tax rate100.0%70.0%
Tax by rate100.0%76.0%
Total100.0%85.3%
Currency100.0%84.0%
Bank account99.3%81.3%

Documents where they differ

Invoice extraction · Czech (150 documents)

GPT-6 Luna is more accurate by 7.5 pp. GPT-6 Luna is 2× cheaper.

MetricGPT-6 LunaQwen3.8 Flash
Correct fields99.3%91.8%
Error-free documents93%77%
Per 1,000 documents$0.51$1.02
To fix / 1,00067233
Speed (median)3.2 sec17.0 sec
Valid answers100.0%96.7%

Accuracy per field

FieldGPT-6 LunaQwen3.8 Flash
Supplier95.3%93.3%
Supplier ID (EIN, IČO…)98.7%94.0%
VAT ID98.0%94.0%
Invoice number100.0%94.7%
Payment reference100.0%94.0%
Invoice date100.0%94.7%
Tax point100.0%94.7%
Due date99.3%94.7%
Net by tax rate100.0%80.0%
Tax by rate100.0%80.7%
Total100.0%94.0%
Currency100.0%94.0%
Bank account99.3%90.7%

Documents where they differ

Invoice extraction · German (150 documents)

GPT-6 Luna is more accurate by 5.7 pp. GPT-6 Luna is 2.1× cheaper.

MetricGPT-6 LunaQwen3.8 Flash
Correct fields99.2%93.5%
Error-free documents91%87%
Per 1,000 documents$0.54$1.12
To fix / 1,00087127
Speed (median)3.7 sec19.0 sec
Valid answers100.0%94.0%

Accuracy per field

FieldGPT-6 LunaQwen3.8 Flash
Supplier100.0%94.0%
Supplier ID (EIN, IČO…)98.7%94.0%
VAT ID99.3%94.0%
Invoice number99.3%92.7%
Payment reference100.0%88.7%
Invoice date98.7%94.0%
Tax point100.0%94.0%
Due date100.0%94.0%
Net by tax rate100.0%94.0%
Tax by rate100.0%94.0%
Total100.0%94.0%
Currency100.0%94.0%
Bank account94.0%94.0%

Documents where they differ

Invoice extraction · Spanish (100 documents)

GPT-6 Luna is more accurate by 11.2 pp. GPT-6 Luna is 2.5× cheaper.

MetricGPT-6 LunaQwen3.8 Flash
Correct fields99.1%87.9%
Error-free documents88%74%
Per 1,000 documents$0.55$1.40
To fix / 1,000120260
Speed (median)3.9 sec22.3 sec
Valid answers100.0%90.0%

Accuracy per field

FieldGPT-6 LunaQwen3.8 Flash
Supplier99.0%86.0%
Supplier ID (EIN, IČO…)100.0%89.0%
VAT ID100.0%86.0%
Invoice number100.0%89.0%
Payment reference98.0%85.0%
Invoice date100.0%89.0%
Tax point100.0%87.0%
Due date100.0%89.0%
Net by tax rate100.0%86.0%
Tax by rate100.0%90.0%
Total100.0%89.0%
Currency100.0%89.0%
Bank account91.0%89.0%

Documents where they differ

Invoice extraction · French (98 documents)

GPT-6 Luna is more accurate by 8.3 pp. GPT-6 Luna is 2.2× cheaper.

MetricGPT-6 LunaQwen3.8 Flash
Correct fields98.9%90.6%
Error-free documents86%82%
Per 1,000 documents$0.57$1.23
To fix / 1,000143180
Speed (median)3.8 sec19.8 sec
Valid answers100.0%95.0%

Accuracy per field

FieldGPT-6 LunaQwen3.8 Flash
Supplier99.0%91.0%
Supplier ID (EIN, IČO…)99.0%91.0%
VAT ID98.0%91.0%
Invoice number99.0%90.0%
Payment reference100.0%86.0%
Invoice date100.0%91.0%
Tax point100.0%92.0%
Due date100.0%89.0%
Net by tax rate100.0%91.0%
Tax by rate100.0%94.0%
Total100.0%91.0%
Currency100.0%91.0%
Bank account90.8%90.0%

Documents where they differ

Email triage · English (US) (150 documents)

Both are equally accurate. GPT-6 Luna is 2.3× cheaper.

MetricGPT-6 LunaQwen3.8 Flash
Correct fields97.5%97.5%
Error-free documents77%79%
Per 1,000 documents$0.13$0.30
To fix / 1,000227213
Speed (median)2.5 sec8.9 sec
Valid answers100.0%100.0%

Accuracy per field

FieldGPT-6 LunaQwen3.8 Flash
Category100.0%100.0%
Priority98.0%97.3%
Sentiment79.3%80.7%
Customer name100.0%100.0%
Order number100.0%100.0%
Invoice number100.0%100.0%
Amount100.0%100.0%
Currency100.0%100.0%
Deadline100.0%99.3%

Documents where they differ

Email triage · Czech (150 documents)

Qwen3.8 Flash is more accurate by 0.6 pp. GPT-6 Luna is 2.5× cheaper.

MetricGPT-6 LunaQwen3.8 Flash
Correct fields97.9%98.5%
Error-free documents81%87%
Per 1,000 documents$0.18$0.45
To fix / 1,000187133
Speed (median)2.6 sec13.0 sec
Valid answers100.0%100.0%

Accuracy per field

FieldGPT-6 LunaQwen3.8 Flash
Category100.0%100.0%
Priority98.7%99.3%
Sentiment82.7%88.7%
Customer name100.0%99.3%
Order number100.0%99.3%
Invoice number100.0%100.0%
Amount100.0%100.0%
Currency100.0%100.0%
Deadline100.0%100.0%

Documents where they differ

Email triage · German (150 documents)

Qwen3.8 Flash is more accurate by 0.9 pp. GPT-6 Luna is 2.6× cheaper.

MetricGPT-6 LunaQwen3.8 Flash
Correct fields97.0%97.9%
Error-free documents73%81%
Per 1,000 documents$0.15$0.39
To fix / 1,000273187
Speed (median)2.8 sec12.1 sec
Valid answers100.0%100.0%

Accuracy per field

FieldGPT-6 LunaQwen3.8 Flash
Category100.0%100.0%
Priority99.3%96.0%
Sentiment73.3%84.7%
Customer name100.0%100.0%
Order number100.0%100.0%
Invoice number100.0%100.0%
Amount100.0%100.0%
Currency100.0%100.0%
Deadline100.0%100.0%

Documents where they differ

Personal data detection · English (US) (150 documents)

GPT-6 Luna is more accurate by 0.4 pp. GPT-6 Luna is 2.6× cheaper.

MetricGPT-6 LunaQwen3.8 Flash
Correct fields99.6%99.2%
Error-free documents97%99%
Per 1,000 documents$0.11$0.28
To fix / 1,0002713
Speed (median)1.8 sec7.5 sec
Valid answers100.0%100.0%

Accuracy per field

FieldGPT-6 LunaQwen3.8 Flash
Names98.7%99.3%
Email addresses100.0%99.3%
Phone numbers100.0%99.3%
Addresses98.7%98.7%
Dates of birth100.0%99.3%
National IDs100.0%99.3%
Bank accounts100.0%99.3%

Documents where they differ

Personal data detection · Czech (150 documents)

GPT-6 Luna is more accurate by 2.7 pp. GPT-6 Luna is 3.4× cheaper.

MetricGPT-6 LunaQwen3.8 Flash
Correct fields99.6%97.0%
Error-free documents97%89%
Per 1,000 documents$0.12$0.41
To fix / 1,00027113
Speed (median)1.8 sec9.6 sec
Valid answers100.0%99.3%

Accuracy per field

FieldGPT-6 LunaQwen3.8 Flash
Names100.0%90.7%
Email addresses100.0%98.0%
Phone numbers100.0%97.3%
Addresses97.3%96.7%
Dates of birth100.0%98.0%
National IDs100.0%99.3%
Bank accounts100.0%98.7%

Documents where they differ

Personal data detection · German (150 documents)

Both are equally accurate. GPT-6 Luna is 3.1× cheaper.

MetricGPT-6 LunaQwen3.8 Flash
Correct fields99.6%99.6%
Error-free documents97%99%
Per 1,000 documents$0.12$0.36
To fix / 1,000277
Speed (median)2.2 sec12.2 sec
Valid answers100.0%100.0%

Accuracy per field

FieldGPT-6 LunaQwen3.8 Flash
Names100.0%99.3%
Email addresses100.0%99.3%
Phone numbers100.0%99.3%
Addresses97.3%99.3%
Dates of birth100.0%100.0%
National IDs100.0%100.0%
Bank accounts100.0%100.0%

Documents where they differ

Contract clauses · English (US) (150 documents)

GPT-6 Luna is more accurate by 4.7 pp. GPT-6 Luna is 3.1× cheaper.

MetricGPT-6 LunaQwen3.8 Flash
Correct fields99.8%95.1%
Error-free documents99%69%
Per 1,000 documents$0.20$0.62
To fix / 1,00013307
Speed (median)2.7 sec14.1 sec
Valid answers100.0%100.0%

Accuracy per field

FieldGPT-6 LunaQwen3.8 Flash
Provider98.7%69.3%
Provider ID100.0%100.0%
Client98.7%69.3%
Client ID100.0%100.0%
Contract type100.0%100.0%
Price (net)100.0%100.0%
Currency100.0%100.0%
Billing period100.0%100.0%
Effective date100.0%99.3%
End date100.0%100.0%
Notice period (months)100.0%100.0%
Late-payment penalty (% per day)100.0%99.3%
Governing law100.0%99.3%

Documents where they differ

Contract clauses · Czech (150 documents)

GPT-6 Luna is more accurate by 0.1 pp. GPT-6 Luna is 2.3× cheaper.

MetricGPT-6 LunaQwen3.8 Flash
Correct fields100.0%99.9%
Error-free documents100%99%
Per 1,000 documents$0.27$0.63
To fix / 1,000013
Speed (median)2.6 sec13.6 sec
Valid answers100.0%100.0%

Accuracy per field

FieldGPT-6 LunaQwen3.8 Flash
Provider100.0%100.0%
Provider ID100.0%100.0%
Client100.0%100.0%
Client ID100.0%100.0%
Contract type100.0%98.7%
Price (net)100.0%100.0%
Currency100.0%100.0%
Billing period100.0%100.0%
Effective date100.0%100.0%
End date100.0%100.0%
Notice period (months)100.0%100.0%
Late-payment penalty (% per day)100.0%100.0%
Governing law100.0%100.0%

Documents where they differ

Contract clauses · German (150 documents)

GPT-6 Luna is more accurate by 2 pp. GPT-6 Luna is 2.9× cheaper.

MetricGPT-6 LunaQwen3.8 Flash
Correct fields100.0%98.0%
Error-free documents100%87%
Per 1,000 documents$0.27$0.78
To fix / 1,0000127
Speed (median)3.0 sec19.7 sec
Valid answers100.0%99.3%

Accuracy per field

FieldGPT-6 LunaQwen3.8 Flash
Provider100.0%99.3%
Provider ID100.0%94.7%
Client100.0%99.3%
Client ID100.0%94.7%
Contract type100.0%99.3%
Price (net)100.0%99.3%
Currency100.0%99.3%
Billing period100.0%99.3%
Effective date100.0%99.3%
End date100.0%99.3%
Notice period (months)100.0%99.3%
Late-payment penalty (% per day)100.0%99.3%
Governing law100.0%91.3%

Documents where they differ