ModelVerdict

Comparison

GPT-6 Luna vs Mistral Large 4

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

Verdict

GPT-6 Luna is more accurate and cheaper in 30 of 30 combinations.

More accurate

GPT-6 Luna

wins 19 of 30

Cheaper

GPT-6 Luna

cheaper in 30 of 30

Faster

GPT-6 Luna

3.0 sec vs 31.5 sec

Fewer fixes

GPT-6 Luna

33 vs 107 documents per 1,000

Across 30 task and document-language combinations (speed and fixes: median).

At a glance

English (US)
Accuracy (correct fields)Per 1,000 documents
Task · documentsGPT-6 LunaMistral Large 4DifferenceGPT-6 LunaMistral Large 4Difference
Invoice extraction99.70%95.39%▼ 4.31$0.51$6.65+1210 %
Email triage97.48%98.74%▲ 1.26$0.13$3.52+2511 %
Personal data detection99.62%99.62%±0$0.11$2.89+2632 %
Contract clauses99.79%96.10%▼ 3.69$0.20$5.24+2507 %
Document summaries100.00%100.00%±0$0.16$5.62+3427 %
Questions about a document99.11%98.89%▼ 0.22$0.15$4.16+2624 %
Czech
Accuracy (correct fields)Per 1,000 documents
Task · documentsGPT-6 LunaMistral Large 4DifferenceGPT-6 LunaMistral Large 4Difference
Invoice extraction99.28%97.08%▼ 2.2$0.51$5.69+1026 %
Email triage97.93%98.15%▲ 0.22$0.18$4.41+2377 %
Personal data detection99.62%99.33%▼ 0.29$0.12$3.15+2538 %
Contract clauses100.00%100.00%±0$0.27$4.58+1596 %
Document summaries100.00%83.33%▼ 16.67$0.21$10.72+4944 %
Questions about a document100.00%89.89%▼ 10.11$0.23$7.08+3016 %
German
Accuracy (correct fields)Per 1,000 documents
Task · documentsGPT-6 LunaMistral Large 4DifferenceGPT-6 LunaMistral Large 4Difference
Invoice extraction99.23%97.49%▼ 1.74$0.54$6.72+1133 %
Email triage96.96%97.85%▲ 0.89$0.15$3.78+2381 %
Personal data detection99.62%99.52%▼ 0.1$0.12$2.61+2124 %
Contract clauses100.00%99.44%▼ 0.56$0.27$5.12+1823 %
Document summaries100.00%99.33%▼ 0.67$0.19$7.19+3609 %
Questions about a document98.00%93.56%▼ 4.44$0.20$5.77+2772 %
Spanish
Accuracy (correct fields)Per 1,000 documents
Task · documentsGPT-6 LunaMistral Large 4DifferenceGPT-6 LunaMistral Large 4Difference
Invoice extraction99.13%96.51%▼ 2.62$0.55$7.49+1255 %
Email triage96.81%97.93%▲ 1.12$0.16$4.03+2455 %
Personal data detection99.71%99.81%▲ 0.1$0.12$3.16+2566 %
Contract clauses100.00%100.00%±0$0.26$4.83+1730 %
Document summaries97.00%91.33%▼ 5.67$0.19$7.99+4133 %
Questions about a document99.67%94.22%▼ 5.45$0.19$5.74+2847 %
French
Accuracy (correct fields)Per 1,000 documents
Task · documentsGPT-6 LunaMistral Large 4DifferenceGPT-6 LunaMistral Large 4Difference
Invoice extraction99.08%88.97%▼ 10.11$0.56$8.71+1461 %
Email triage97.26%98.30%▲ 1.04$0.16$3.70+2211 %
Personal data detection99.62%98.67%▼ 0.95$0.12$2.96+2282 %
Contract clauses99.33%99.28%▼ 0.05$0.27$4.60+1584 %
Document summaries98.78%72.67%▼ 26.11$0.20$11.26+5632 %
Questions about a document99.56%92.33%▼ 7.23$0.20$5.74+2712 %

Difference: Mistral Large 4 vs. GPT-6 Luna; accuracy in percentage points, price in percent. The better value is in bold. Click a task to see the details.

Invoice extraction · English (US) (150 documents)

GPT-6 Luna is more accurate by 4.3 pp. GPT-6 Luna is 13.1× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields99.70%95.39%
Error-free documents96.67%77.33%
Per 1,000 documents$0.51$6.65
To fix / 1,00033227
Speed (median)3.4 sec29.0 sec
Valid answers100.00%97.33%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Invoice extraction · Czech (150 documents)

GPT-6 Luna is more accurate by 2.2 pp. GPT-6 Luna is 11.3× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields99.28%97.08%
Error-free documents93.33%88.67%
Per 1,000 documents$0.51$5.69
To fix / 1,00067113
Speed (median)3.2 sec26.3 sec
Valid answers100.00%98.00%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Invoice extraction · German (150 documents)

GPT-6 Luna is more accurate by 1.7 pp. GPT-6 Luna is 12.3× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields99.23%97.49%
Error-free documents91.33%91.33%
Per 1,000 documents$0.54$6.72
To fix / 1,0008787
Speed (median)3.7 sec35.3 sec
Valid answers100.00%98.00%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Invoice extraction · Spanish (150 documents)

GPT-6 Luna is more accurate by 2.6 pp. GPT-6 Luna is 13.5× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields99.13%96.51%
Error-free documents88.67%88.00%
Per 1,000 documents$0.55$7.49
To fix / 1,000113120
Speed (median)3.9 sec41.3 sec
Valid answers100.00%97.33%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Invoice extraction · French (150 documents)

GPT-6 Luna is more accurate by 10.1 pp. GPT-6 Luna is 15.6× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields99.08%88.97%
Error-free documents88.00%65.33%
Per 1,000 documents$0.56$8.71
To fix / 1,000120347
Speed (median)3.9 sec43.4 sec
Valid answers100.00%91.33%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Email triage · English (US) (150 documents)

Mistral Large 4 is more accurate by 1.3 pp. GPT-6 Luna is 26.1× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields97.48%98.74%
Error-free documents77.33%88.67%
Per 1,000 documents$0.13$3.52
To fix / 1,000227113
Speed (median)2.5 sec19.8 sec
Valid answers100.00%100.00%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Email triage · Czech (150 documents)

Mistral Large 4 is more accurate by 0.2 pp. GPT-6 Luna is 24.8× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields97.93%98.15%
Error-free documents81.33%89.33%
Per 1,000 documents$0.18$4.41
To fix / 1,000187107
Speed (median)2.6 sec28.4 sec
Valid answers100.00%99.33%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Email triage · German (150 documents)

Mistral Large 4 is more accurate by 0.9 pp. GPT-6 Luna is 24.8× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields96.96%97.85%
Error-free documents72.67%82.00%
Per 1,000 documents$0.15$3.78
To fix / 1,000273180
Speed (median)2.8 sec21.7 sec
Valid answers100.00%100.00%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Email triage · Spanish (150 documents)

Mistral Large 4 is more accurate by 1.1 pp. GPT-6 Luna is 25.5× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields96.81%97.93%
Error-free documents72.00%82.00%
Per 1,000 documents$0.16$4.03
To fix / 1,000280180
Speed (median)3.1 sec25.4 sec
Valid answers100.00%100.00%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Email triage · French (150 documents)

Mistral Large 4 is more accurate by 1 pp. GPT-6 Luna is 23.1× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields97.26%98.30%
Error-free documents76.00%84.67%
Per 1,000 documents$0.16$3.70
To fix / 1,000240153
Speed (median)2.8 sec24.5 sec
Valid answers100.00%100.00%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

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

Both are equally accurate. GPT-6 Luna is 27.3× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields99.62%99.62%
Error-free documents97.33%97.33%
Per 1,000 documents$0.11$2.89
To fix / 1,0002727
Speed (median)1.8 sec28.2 sec
Valid answers100.00%100.00%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Personal data detection · Czech (150 documents)

GPT-6 Luna is more accurate by 0.3 pp. GPT-6 Luna is 26.4× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields99.62%99.33%
Error-free documents97.33%95.33%
Per 1,000 documents$0.12$3.15
To fix / 1,0002747
Speed (median)1.8 sec18.5 sec
Valid answers100.00%100.00%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Personal data detection · German (150 documents)

GPT-6 Luna is more accurate by 0.1 pp. GPT-6 Luna is 22.2× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields99.62%99.52%
Error-free documents97.33%96.67%
Per 1,000 documents$0.12$2.61
To fix / 1,0002733
Speed (median)2.2 sec18.6 sec
Valid answers100.00%100.00%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Personal data detection · Spanish (150 documents)

Mistral Large 4 is more accurate by 0.1 pp. GPT-6 Luna is 26.7× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields99.71%99.81%
Error-free documents98.00%98.67%
Per 1,000 documents$0.12$3.16
To fix / 1,0002013
Speed (median)2.3 sec21.9 sec
Valid answers100.00%100.00%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Personal data detection · French (150 documents)

GPT-6 Luna is more accurate by 0.9 pp. GPT-6 Luna is 23.8× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields99.62%98.67%
Error-free documents97.33%94.67%
Per 1,000 documents$0.12$2.96
To fix / 1,0002753
Speed (median)2.4 sec23.6 sec
Valid answers100.00%99.33%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Contract clauses · English (US) (150 documents)

GPT-6 Luna is more accurate by 3.7 pp. GPT-6 Luna is 26.1× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields99.79%96.10%
Error-free documents98.67%74.67%
Per 1,000 documents$0.20$5.24
To fix / 1,00013253
Speed (median)2.7 sec41.4 sec
Valid answers100.00%100.00%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Contract clauses · Czech (150 documents)

Both are equally accurate. GPT-6 Luna is 17× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields100.00%100.00%
Error-free documents100.00%100.00%
Per 1,000 documents$0.27$4.58
To fix / 1,00000
Speed (median)2.6 sec43.9 sec
Valid answers100.00%100.00%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Contract clauses · German (150 documents)

GPT-6 Luna is more accurate by 0.6 pp. GPT-6 Luna is 19.2× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields100.00%99.44%
Error-free documents100.00%95.33%
Per 1,000 documents$0.27$5.12
To fix / 1,000047
Speed (median)3.0 sec36.2 sec
Valid answers100.00%100.00%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Contract clauses · Spanish (150 documents)

Both are equally accurate. GPT-6 Luna is 18.3× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields100.00%100.00%
Error-free documents100.00%100.00%
Per 1,000 documents$0.26$4.83
To fix / 1,00000
Speed (median)3.1 sec35.0 sec
Valid answers100.00%100.00%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Contract clauses · French (150 documents)

Both are equally accurate. GPT-6 Luna is 16.8× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields99.33%99.28%
Error-free documents91.33%90.67%
Per 1,000 documents$0.27$4.60
To fix / 1,0008793
Speed (median)3.2 sec34.7 sec
Valid answers100.00%100.00%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Document summaries · English (US) (150 documents)

Both are equally accurate. GPT-6 Luna is 35.3× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields100.00%100.00%
Error-free documents100.00%100.00%
Per 1,000 documents$0.16$5.62
To fix / 1,00000
Speed (median)2.9 sec37.2 sec
Valid answers100.00%100.00%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Document summaries · Czech (150 documents)

GPT-6 Luna is more accurate by 16.7 pp. GPT-6 Luna is 50.4× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields100.00%83.33%
Error-free documents100.00%83.33%
Per 1,000 documents$0.21$10.72
To fix / 1,0000167
Speed (median)3.4 sec55.4 sec
Valid answers100.00%83.33%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Document summaries · German (150 documents)

GPT-6 Luna is more accurate by 0.7 pp. GPT-6 Luna is 37.1× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields100.00%99.33%
Error-free documents100.00%99.33%
Per 1,000 documents$0.19$7.19
To fix / 1,00007
Speed (median)3.3 sec48.0 sec
Valid answers100.00%99.33%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Document summaries · Spanish (150 documents)

GPT-6 Luna is more accurate by 5.7 pp. GPT-6 Luna is 42.3× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields97.00%91.33%
Error-free documents82.00%91.33%
Per 1,000 documents$0.19$7.99
To fix / 1,00018087
Speed (median)3.0 sec38.3 sec
Valid answers100.00%91.33%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Document summaries · French (150 documents)

GPT-6 Luna is more accurate by 26.1 pp. GPT-6 Luna is 57.3× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields98.78%72.67%
Error-free documents92.67%72.67%
Per 1,000 documents$0.20$11.26
To fix / 1,00073273
Speed (median)3.2 sec46.4 sec
Valid answers100.00%72.67%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Questions about a document · English (US) (150 documents)

GPT-6 Luna is more accurate by 0.2 pp. GPT-6 Luna is 27.2× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields99.11%98.89%
Error-free documents96.00%97.33%
Per 1,000 documents$0.15$4.16
To fix / 1,0004027
Speed (median)3.1 sec17.7 sec
Valid answers100.00%99.33%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Questions about a document · Czech (150 documents)

GPT-6 Luna is more accurate by 10.1 pp. GPT-6 Luna is 31.2× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields100.00%89.89%
Error-free documents100.00%78.00%
Per 1,000 documents$0.23$7.08
To fix / 1,0000220
Speed (median)3.1 sec33.6 sec
Valid answers100.00%93.33%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Questions about a document · German (150 documents)

GPT-6 Luna is more accurate by 4.4 pp. GPT-6 Luna is 28.7× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields98.00%93.56%
Error-free documents91.33%89.33%
Per 1,000 documents$0.20$5.77
To fix / 1,00087107
Speed (median)3.2 sec31.5 sec
Valid answers100.00%94.67%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Questions about a document · Spanish (150 documents)

GPT-6 Luna is more accurate by 5.4 pp. GPT-6 Luna is 29.5× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields99.67%94.22%
Error-free documents98.67%90.67%
Per 1,000 documents$0.19$5.74
To fix / 1,0001393
Speed (median)2.9 sec22.8 sec
Valid answers100.00%95.33%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Questions about a document · French (150 documents)

GPT-6 Luna is more accurate by 7.2 pp. GPT-6 Luna is 28.1× cheaper.

MetricGPT-6 LunaMistral Large 4
Correct fields99.56%92.33%
Error-free documents98.00%87.33%
Per 1,000 documents$0.20$5.74
To fix / 1,00020127
Speed (median)3.0 sec19.8 sec
Valid answers100.00%94.00%

Free with login

Accuracy field by field and the documents where the models differ.

Just your email, no password.

Log in with email

We use your email only to log you in (and for the digest if you tick it). No spam.