ModelVerdict

Comparison

Mistral Medium 3.5 vs Qwen3.8 Flash

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

Verdict

Mistral Medium 3.5 is more accurate; Qwen3.8 Flash is cheaper in 30 of 30 combinations.

More accurate

Mistral Medium 3.5

wins 21 of 30

Cheaper

Qwen3.8 Flash

cheaper in 30 of 30

Faster

Mistral Medium 3.5

1.0 sec vs 14.1 sec

Fewer fixes

Mistral Medium 3.5

20 vs 67 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 · documentsMistral Medium 3.5Qwen3.8 FlashDifferenceMistral Medium 3.5Qwen3.8 FlashDifference
Invoice extraction97.76%82.36%▼ 15.4$4.37$1.15−74 %
Email triage94.89%97.48%▲ 2.59$1.47$0.30−79 %
Personal data detection99.71%99.24%▼ 0.47$1.15$0.28−76 %
Contract clauses98.10%95.13%▼ 2.97$2.35$0.62−74 %
Document summaries100.00%99.89%▼ 0.11$1.57$0.75−52 %
Questions about a document100.00%99.78%▼ 0.22$1.81$0.44−76 %
Czech
Accuracy (correct fields)Per 1,000 documents
Task · documentsMistral Medium 3.5Qwen3.8 FlashDifferenceMistral Medium 3.5Qwen3.8 FlashDifference
Invoice extraction93.49%91.79%▼ 1.7$4.69$1.02−78 %
Email triage94.52%98.52%▲ 4$1.95$0.45−77 %
Personal data detection99.81%96.95%▼ 2.86$1.40$0.41−71 %
Contract clauses100.00%99.90%▼ 0.1$3.06$0.63−79 %
Document summaries100.00%95.44%▼ 4.56$2.31$1.48−36 %
Questions about a document100.00%94.22%▼ 5.78$2.51$0.71−72 %
German
Accuracy (correct fields)Per 1,000 documents
Task · documentsMistral Medium 3.5Qwen3.8 FlashDifferenceMistral Medium 3.5Qwen3.8 FlashDifference
Invoice extraction97.38%93.49%▼ 3.89$4.72$1.12−76 %
Email triage92.74%97.85%▲ 5.11$1.77$0.39−78 %
Personal data detection100.00%99.62%▼ 0.38$1.38$0.36−74 %
Contract clauses97.13%98.00%▲ 0.87$2.76$0.78−72 %
Document summaries100.00%95.33%▼ 4.67$1.98$1.43−28 %
Questions about a document100.00%98.89%▼ 1.11$2.13$0.46−78 %
Spanish
Accuracy (correct fields)Per 1,000 documents
Task · documentsMistral Medium 3.5Qwen3.8 FlashDifferenceMistral Medium 3.5Qwen3.8 FlashDifference
Invoice extraction96.87%88.87%▼ 8$4.75$1.38−71 %
Email triage94.15%96.74%▲ 2.59$1.79$0.46−74 %
Personal data detection99.62%99.81%▲ 0.19$1.41$0.38−73 %
Contract clauses100.00%100.00%±0$3.02$0.60−80 %
Document summaries100.00%94.44%▼ 5.56$1.86$1.22−34 %
Questions about a document100.00%99.67%▼ 0.33$2.18$0.53−76 %
French
Accuracy (correct fields)Per 1,000 documents
Task · documentsMistral Medium 3.5Qwen3.8 FlashDifferenceMistral Medium 3.5Qwen3.8 FlashDifference
Invoice extraction98.56%90.46%▼ 8.1$4.82$1.28−73 %
Email triage94.15%96.44%▲ 2.29$1.78$0.45−75 %
Personal data detection99.62%99.71%▲ 0.09$1.43$0.40−72 %
Contract clauses100.00%99.49%▼ 0.51$3.01$0.66−78 %
Document summaries100.00%96.33%▼ 3.67$1.87$1.46−22 %
Questions about a document100.00%98.78%▼ 1.22$2.19$0.57−74 %

Difference: Qwen3.8 Flash vs. Mistral Medium 3.5; 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)

Mistral Medium 3.5 is more accurate by 15.4 pp. Qwen3.8 Flash is 3.8× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields97.76%82.36%
Error-free documents79.33%57.33%
Per 1,000 documents$4.37$1.15
To fix / 1,000207427
Speed (median)2.3 sec17.4 sec
Valid answers100.00%90.67%

Free with login

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

Just your email, no password.

Invoice extraction · Czech (150 documents)

Mistral Medium 3.5 is more accurate by 1.7 pp. Qwen3.8 Flash is 4.6× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields93.49%91.79%
Error-free documents63.33%76.67%
Per 1,000 documents$4.69$1.02
To fix / 1,000367233
Speed (median)2.6 sec17.0 sec
Valid answers100.00%96.67%

Free with login

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

Just your email, no password.

Invoice extraction · German (150 documents)

Mistral Medium 3.5 is more accurate by 3.9 pp. Qwen3.8 Flash is 4.2× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields97.38%93.49%
Error-free documents74.67%87.33%
Per 1,000 documents$4.72$1.12
To fix / 1,000253127
Speed (median)2.5 sec19.0 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.

Invoice extraction · Spanish (150 documents)

Mistral Medium 3.5 is more accurate by 8 pp. Qwen3.8 Flash is 3.4× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields96.87%88.87%
Error-free documents66.67%74.67%
Per 1,000 documents$4.75$1.38
To fix / 1,000333253
Speed (median)2.5 sec22.0 sec
Valid answers100.00%90.67%

Free with login

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

Just your email, no password.

Invoice extraction · French (150 documents)

Mistral Medium 3.5 is more accurate by 8.1 pp. Qwen3.8 Flash is 3.8× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields98.56%90.46%
Error-free documents84.67%83.33%
Per 1,000 documents$4.82$1.28
To fix / 1,000153167
Speed (median)2.6 sec19.4 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.

Email triage · English (US) (150 documents)

Qwen3.8 Flash is more accurate by 2.6 pp. Qwen3.8 Flash is 4.8× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields94.89%97.48%
Error-free documents57.33%78.67%
Per 1,000 documents$1.47$0.30
To fix / 1,000427213
Speed (median)0.8 sec8.9 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.

Email triage · Czech (150 documents)

Qwen3.8 Flash is more accurate by 4 pp. Qwen3.8 Flash is 4.4× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields94.52%98.52%
Error-free documents55.33%86.67%
Per 1,000 documents$1.95$0.45
To fix / 1,000447133
Speed (median)0.8 sec13.0 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.

Email triage · German (150 documents)

Qwen3.8 Flash is more accurate by 5.1 pp. Qwen3.8 Flash is 4.5× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields92.74%97.85%
Error-free documents48.00%81.33%
Per 1,000 documents$1.77$0.39
To fix / 1,000520187
Speed (median)0.8 sec12.1 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)

Qwen3.8 Flash is more accurate by 2.6 pp. Qwen3.8 Flash is 3.9× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields94.15%96.74%
Error-free documents54.00%72.00%
Per 1,000 documents$1.79$0.46
To fix / 1,000460280
Speed (median)0.9 sec11.2 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.

Email triage · French (150 documents)

Qwen3.8 Flash is more accurate by 2.3 pp. Qwen3.8 Flash is 3.9× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields94.15%96.44%
Error-free documents54.00%76.00%
Per 1,000 documents$1.78$0.45
To fix / 1,000460240
Speed (median)0.9 sec13.2 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.

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

Mistral Medium 3.5 is more accurate by 0.5 pp. Qwen3.8 Flash is 4.1× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields99.71%99.24%
Error-free documents98.00%98.67%
Per 1,000 documents$1.15$0.28
To fix / 1,0002013
Speed (median)0.8 sec7.5 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 · Czech (150 documents)

Mistral Medium 3.5 is more accurate by 2.9 pp. Qwen3.8 Flash is 3.4× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields99.81%96.95%
Error-free documents98.67%88.67%
Per 1,000 documents$1.40$0.41
To fix / 1,00013113
Speed (median)0.8 sec9.6 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.

Personal data detection · German (150 documents)

Mistral Medium 3.5 is more accurate by 0.4 pp. Qwen3.8 Flash is 3.9× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields100.00%99.62%
Error-free documents100.00%99.33%
Per 1,000 documents$1.38$0.36
To fix / 1,00007
Speed (median)0.9 sec12.2 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 · Spanish (150 documents)

Qwen3.8 Flash is more accurate by 0.2 pp. Qwen3.8 Flash is 3.7× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields99.62%99.81%
Error-free documents97.33%98.67%
Per 1,000 documents$1.41$0.38
To fix / 1,0002713
Speed (median)1.0 sec10.2 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 · French (150 documents)

Qwen3.8 Flash is more accurate by 0.1 pp. Qwen3.8 Flash is 3.6× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields99.62%99.71%
Error-free documents97.33%98.00%
Per 1,000 documents$1.43$0.40
To fix / 1,0002720
Speed (median)0.9 sec10.9 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 · English (US) (150 documents)

Mistral Medium 3.5 is more accurate by 3 pp. Qwen3.8 Flash is 3.8× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields98.10%95.13%
Error-free documents87.33%69.33%
Per 1,000 documents$2.35$0.62
To fix / 1,000127307
Speed (median)1.1 sec14.1 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)

Mistral Medium 3.5 is more accurate by 0.1 pp. Qwen3.8 Flash is 4.9× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields100.00%99.90%
Error-free documents100.00%98.67%
Per 1,000 documents$3.06$0.63
To fix / 1,000013
Speed (median)1.2 sec13.6 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.

Contract clauses · German (150 documents)

Qwen3.8 Flash is more accurate by 0.9 pp. Qwen3.8 Flash is 3.5× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields97.13%98.00%
Error-free documents75.33%87.33%
Per 1,000 documents$2.76$0.78
To fix / 1,000247127
Speed (median)1.2 sec19.7 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.

Contract clauses · Spanish (150 documents)

Both are equally accurate. Qwen3.8 Flash is 5.1× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields100.00%100.00%
Error-free documents100.00%100.00%
Per 1,000 documents$3.02$0.60
To fix / 1,00000
Speed (median)1.3 sec14.2 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 · French (150 documents)

Mistral Medium 3.5 is more accurate by 0.5 pp. Qwen3.8 Flash is 4.6× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields100.00%99.49%
Error-free documents100.00%93.33%
Per 1,000 documents$3.01$0.66
To fix / 1,000067
Speed (median)1.3 sec16.2 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 · English (US) (150 documents)

Mistral Medium 3.5 is more accurate by 0.1 pp. Qwen3.8 Flash is 2.1× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields100.00%99.89%
Error-free documents100.00%99.33%
Per 1,000 documents$1.57$0.75
To fix / 1,00007
Speed (median)0.9 sec17.8 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.

Document summaries · Czech (150 documents)

Mistral Medium 3.5 is more accurate by 4.6 pp. Qwen3.8 Flash is 1.6× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields100.00%95.44%
Error-free documents100.00%93.33%
Per 1,000 documents$2.31$1.48
To fix / 1,000067
Speed (median)1.4 sec34.2 sec
Valid answers100.00%96.00%

Free with login

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

Just your email, no password.

Document summaries · German (150 documents)

Mistral Medium 3.5 is more accurate by 4.7 pp. Qwen3.8 Flash is 1.4× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields100.00%95.33%
Error-free documents100.00%95.33%
Per 1,000 documents$1.98$1.43
To fix / 1,000047
Speed (median)1.1 sec29.5 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.

Document summaries · Spanish (150 documents)

Mistral Medium 3.5 is more accurate by 5.6 pp. Qwen3.8 Flash is 1.5× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields100.00%94.44%
Error-free documents100.00%93.33%
Per 1,000 documents$1.86$1.22
To fix / 1,000067
Speed (median)1.0 sec23.9 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.

Document summaries · French (150 documents)

Mistral Medium 3.5 is more accurate by 3.7 pp. Qwen3.8 Flash is 1.3× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields100.00%96.33%
Error-free documents100.00%94.67%
Per 1,000 documents$1.87$1.46
To fix / 1,000053
Speed (median)1.0 sec28.1 sec
Valid answers100.00%96.67%

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)

Mistral Medium 3.5 is more accurate by 0.2 pp. Qwen3.8 Flash is 4.1× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields100.00%99.78%
Error-free documents100.00%98.67%
Per 1,000 documents$1.81$0.44
To fix / 1,000013
Speed (median)0.8 sec8.7 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.

Questions about a document · Czech (150 documents)

Mistral Medium 3.5 is more accurate by 5.8 pp. Qwen3.8 Flash is 3.5× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields100.00%94.22%
Error-free documents100.00%84.67%
Per 1,000 documents$2.51$0.71
To fix / 1,0000153
Speed (median)0.9 sec14.3 sec
Valid answers100.00%98.67%

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)

Mistral Medium 3.5 is more accurate by 1.1 pp. Qwen3.8 Flash is 4.6× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields100.00%98.89%
Error-free documents100.00%96.67%
Per 1,000 documents$2.13$0.46
To fix / 1,000033
Speed (median)0.8 sec8.6 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.

Questions about a document · Spanish (150 documents)

Mistral Medium 3.5 is more accurate by 0.3 pp. Qwen3.8 Flash is 4.2× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields100.00%99.67%
Error-free documents100.00%98.67%
Per 1,000 documents$2.18$0.53
To fix / 1,000013
Speed (median)0.8 sec10.3 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.

Questions about a document · French (150 documents)

Mistral Medium 3.5 is more accurate by 1.2 pp. Qwen3.8 Flash is 3.9× cheaper.

MetricMistral Medium 3.5Qwen3.8 Flash
Correct fields100.00%98.78%
Error-free documents100.00%97.33%
Per 1,000 documents$2.19$0.57
To fix / 1,000027
Speed (median)0.9 sec10.4 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.

Log in with email

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